Content uploaded by Edmund John Fordham
Author content
All content in this area was uploaded by Edmund John Fordham on Jan 19, 2023
Content may be subject to copyright.
Content uploaded by Edmund John Fordham
Author content
All content in this area was uploaded by Edmund John Fordham on Jul 15, 2022
Content may be subject to copyright.
Content uploaded by Edmund John Fordham
Author content
All content in this area was uploaded by Edmund John Fordham on Jun 28, 2022
Content may be subject to copyright.
Content uploaded by Edmund John Fordham
Author content
All content in this area was uploaded by Edmund John Fordham on Mar 13, 2022
Content may be subject to copyright.
1
Hazardous Substances potentially generated in “loss of control” accidents in Li-ion
Battery Energy Storage Systems (BESS) :
storage capacities implying Hazardous Substances Consent obligations
EurIng Dr Edmund Fordham MA PhD CPhys CEng FInstP
Professor Sir David Melville CBE CPhys FInstP
Abstract
An outline is given of the primary legislation and the defining Regulations that determine a legal
obligation to seek Hazardous Substances Consent (HSC) from the local Planning Authority before installing
plant that contains Hazardous Substances (HS) above specified thresholds. The specific provisions and
sources relate to installations in England but parallel legislation covers other UK nations and in fact across the
EU. A provision found in “Part 3” (Substances Used in Processes) is frequently overlooked. This provides that
where it is “reasonable to foresee” Hazardous Substances specified in Parts 1 or 2 being generated “if control
of the processes is lost”, then any “Substance(s) S” that are “used in the process” are to be regarded as
Hazardous Substances. The Controlled Quantities of such “Substance(s) S” are those amounts which it is
believed may generate (in loss of control accidents) quantities exceeding the Controlled Quantities of the
Hazardous Substances listed in Parts 1 or 2.
Li-ion BESS are well-known to generate in “loss of control” accidents (popularly if incorrectly known as
“battery fires”) several Hazardous Substances listed in Parts 1 or 2. For application of the HSC Regulations to
Li-ion BESS one therefore requires a technically sound determination of the Controlled Quantities of the
“Substance(s) S” (i.e. battery constituent chemicals) that would oblige the operator to seek HSC.
In the absence of actual closed-container and open fire tests of representative samples of the actual
battery cells to be installed, this paper uses literature sources to make estimates of the Controlled Quantities,
based on stated energy storage capacity in MWh, since detailed chemical composition is rarely declared.
We analyse the failure behaviour of Lithium Iron Phosphate (LFP) cells versus the qualitatively different
“mixed oxide” cell types (NMC, LCO etc), and criteria for HSC obligations based on P2 Flammable Gases and
H1 and H2 Acute Toxic Gases, and other considerations such as the explosive failure potential of highly
charged cells. We use this to estimate threshold energy storage capacities likely to carry a HSC obligation.
Also explored are: possible classification as Explosive Articles; possible generation of inhalable Nickel
Oxides in fire; and possible generation of E1 Environmental Hazard in contaminated firewater.
Tab le 13 makes clear that most HSC thresholds are well below 50 MWh. Hence Li-ion BESS of capacity 50
MWh, regardless of cell type, will almost inevitably carry a HSC obligation. We also show that if Li-ion BESS
cells are regarded as Explosive Articles, the threshold is much lower, about 1.5 MWh. Many such Li-ion BESS
are currently being installed under Planning Consents with rated powers just below 50 MW (and typical
storage somewhat above 50 MWh). In such cases, unless HSC has been applied for and granted, it would
appear that an offence under S. 23 of the Planning (Hazardous Substances) Act 1990 has occurred.
For those grid-scale BESS with storage of several hundred MWh which are currently planned, the
requirement for HSC becomes indisputable.
Version 3.9 17 January 2023
A typographical error on page 32 has been corrected, with
consequent changes to Table 13 on page 50 and in the summaries.
2
Table of Contents
Executive Summary p 4
1.0 The Planning (Hazardous Substances) Act 1990 p 7
1.1 The Planning (Hazardous Substances) Regulations 2015 p 8
1.2 Hazardous Substances and Controlled Quantities under Part 3: p 10
1.3 The “Aggregation Rule” of Note 5 p 11
1.4 Other relevant Notes in Part 4 p 12
2.0 Application of the HSC Regulations to BESS p 13
2.1 Determination of CQ of integral substance(s) S hazardous under Part 3 p 13
2.2 Relationship of the HSC Regulations to the COMAH Regulations p 15
3.0 Hazardous Substances under Parts 1 or 2 generated in loss of control in BESS p 16
3.1 Routes to generation of Part 1,2 Hazardous Substances in loss of control
accidents in BESS:
(I) Generation of P2 Flammable Gases in anoxic conditions p 20
3.2 Other Physical Hazards: P1a and P1b Explosives p 23
3.2.1 Field tests by Professor Christensen, Newcastle University p 27
3.2.2 Charged cells as Explosive Articles p 30
3.3 Routes to generation of Part 1,2 Hazardous Substances in loss of control
accidents in BESS:
(II) Generation of H1 and H2 Acute Toxic gases in oxidising conditions p 32
3.4 Routes to generation of Part 1,2 Hazardous Substances in loss of control
accidents in BESS:
(III) inhalable Nickel compounds p 39
3.4.1 Evidence for the generation of inhalable Nickel Oxides
in loss of control accidents p 40
3.5 Routes to generation of Part 1,2 Hazardous Substances in loss of control
accidents in BESS:
(IV) Generation of E1 Hazards to the Aquatic Environment in firefighting
operations p 47
4.0 Conclusions p 49
3
List of Tables
1. Gaseous Hazardous Substances generated in BESS loss of control accidents p 19
2. Environmentally Hazardous Substances potentially generated from electrode materials in BESS loss of
control accidents p 20
3. Energy densities, thermal runaway temperatures, and total gas generation in anoxic conditions, from
two contrasting cell types in thermal runaway p 22
4. Published gas mole fractions, and implied masses per unit energy storage, from two contrasting cell
types in thermal runaway p 22
5. Calculated gas masses from Table 4; the same data in tonnes/50 MWh; and contributions to the sum
required in the Aggregation Rule (for a 50 MWh BESS). p 22
6. Fluoride gases generated from LFP cells compared to LCO or NMC cells from two sources. p 33
7. Tonnages of active substances at an actual 26.3 MWh BESS in Northern Ireland p 34
8. Estimated carbon content of Li-ion cells used in the BESS in Table 7. p 35
9. CO2/CO ratios in open fire tests reported by F M Global and application to Table 7 BESS p 35
10. Mass of CO generated from carbon-containing components of a 26.3 MWh BESS, under two
assumptions for mole fraction of CO in combustion products. p 36
11. Toxic gases generated from BESS in oxidising conditions for two contrasting cell types and a range of
quantities of CO generated from carbon-containing components. p 38
12. Specified Nickel compounds that are Part 2 Named Hazardous Substances if present in “inhalable
powder form”, being known or suspected carcinogens by inhalation p 39
13. Summary of energy storage capacity thresholds below 50 MWh likely to trigger a requirement for HSC,
for contrasting cathode chemistries and CO assumptions. p 50
List of Figures
1. (a) A charged 2.6 Ah cell being deliberately overheated.
(b) at the point of rupture
(c) the cell takes off as a rocket
(d) seconds later the discharge is complete, and the cell is red hot. p 24
2. (a) Stack of fully-charged pouch cells in fire bunker for thermal runaway tests.
(b) The first cell fails explosively in a white-hot fireball at 0:09/2:13 minutes.
(c) subsequent jets of flame last another 8 seconds
(d) a second cell goes into thermal runaway with a similar fireball, lasting about 5 seconds
(e) the last of 5 or 6 such fireballs p 25-26
3. (a) A 23 kg hammer and nail begins to fall on the EV Li-ion module
(b) The hammer at the instant of penetration
(c) The hammer of 23 kg weight is thrown back into the air as the cells disrupt explosively
(d) A fraction of a second later, a black cloud of “cathode material” emerges
(e) Ignition of flammable gases follows, with a large fireball emitting considerable radiant heat
(f) The “long, flare-like flames” last for several minutes as the fire progresses p 27-29
4. Summary of possible courses of events in a BESS “thermal runaway” event. p 40
5. (a) As Figure 3(d), showing the early emergence of a “black cloud” before a “white cloud”
(b) Initial failure (with black cloud) of a 1.67 kWh module in a low State of Charge (SoC) (40%)
(c) The white vapour cloud follows a fraction of a second later.
(d) The vapour cloud eventually fills the container, without ignition, during the test. p 41-42
6. (a) From internal CCTV, seconds before the Liverpool explosion.
(b) The vapour cloud fills the local region of the container before ignition. p 43
4
Executive Summary
1. The Planning (Hazardous Substances) Act 1990 established the District Council (in England)
as the Hazardous Substances Authority (HSA). Hazardous Substances (HS, actual, planned or
foreseeable) “in, on or over the land” above specified thresholds, require the consent of the
HSA. It is an offence to introduce Hazardous Substances (HS) above such thresholds without
Hazardous Substances Consent (HSC) being applied for and granted. Persons controlling the
land, knowingly introducing Hazardous Substances (HS), or allowing HS to be present, can all
be held liable for the offence. The HSA also has the power to issue a Hazardous Substances
Contravention Notice.
2. The definition of a Hazardous Substance (HS), and procedural matters, are provided in the
Planning (Hazardous Substances) Regulations 2015, referred to here as the HSC Regulations.
The related COMAH (Control Of Major Accident Hazards) Regulations are the UK companion
to the HSC Regulations from the perspective of operational safety, in compliance with the
Seveso III Directive. The COMAH Regulations refer to “Dangerous Substances” but the
definitions are identical. Both Regulations use definitions of Hazardous Substances (HS)
prescribed by the Seveso III Directive of the EU, and remain in force in the UK
implementations.
3. Hazardous Substances HS are defined in the HSC Regulations under Part 1 (Categories of
hazard – inflammables, toxics, etc), Part 2 (named substances) and Part 3 (“Substances used
in processes”) which applies “where it is reasonable to foresee that a HS falling within Part 1
or Part 2 may be generated during loss of control of the processes”. In such circumstances,
any Substance “S” whatever “used in the process” can be held to be a “Hazardous
Substance”. Part 3 is frequently overlooked, yet it is when “control of the processes is lost”
that a Li-ion BESS will generate several Hazardous Substances listed in Parts 1 and 2. A
“thermal runaway” or battery “fire” in a BESS represents “loss of control of the processes”.
4. Note 6 to the HSC Regulations makes clear that even Hazardous Substances not covered by
the CLP [ Classification Labelling and Packaging ] Regulation (which includes waste, and
substances contained within “articles”) must be assigned to “the most analogous category
falling in the scope of the Regulations” if they “possess, or are likely to possess … equivalent
properties in terms of major accident potential”.
5. The current position of the Health and Safety Executive (HSE) is that the COMAH and HSC
Regulations do not apply to grid-scale BESS because batteries are “articles” under the CLP
Regulation. Note 6 makes clear however that being “not covered by the CLP” does not
amount to an exemption. Enquiry of HSE has failed to cite any legal provision that provides
an exemption for batteries. A companion paper deals with the legal interpretations, but the
position of HSE appears to be wrong in law.
6. It is the mere presence of Hazardous Substances beyond the Controlled Quantities (CQs)
that leads to the requirement to seek HSC. Note 4 makes clear that it is the total inventory
present at an “establishment” that must be considered in relation to the Controlled Quantity
(CQ). Spatial arrangement of the plant is irrelevant. Nor is the likelihood of accidents
happening, or of propagating, a relevant consideration. For Part 3 to apply, all that is
required is that it must be “reasonable to foresee” the generation of HS listed in Parts 1 and
2 “if control of the processes is lost”, and this is now beyond speculation: it is known fact.
5
7. Hazardous Substances (HS) listed in Parts 1 and 2 that are known to be generated in Li-ion
BESS accidents include: P2 Flammable Gases (Part 1) including methane, ethylene, ethane
and carbon monoxide; H1 and H2 Acute Toxic gases (Part 1), including hydrogen cyanide,
carbon monoxide, hydrogen fluoride and other unstable toxic fluorides. Explosives hazards
explicitly cover “explosive articles” and Li-ion cells can certainly fail explosively and may
qualify as Division 1.3 “explosive articles” and thus as a P1a Explosive (Part 1). Hydrogen may
also be generated and is a Named Substance under Part 2.
8. The Controlled Quantities (CQs) of substance(s) S under Part 3 (i.e. battery component
chemicals) are: ”The amount of S which it is believed may generate (on its own or in
combination with other substances used in the relevant process) an amount equal to or
exceeding the controlled quantity of the HS in question.” Hence it is necessary to conduct a
competent engineering analysis to determine, in failure situations, how much of “S” is
needed to generate CQs of the various HS listed in Parts 1 and 2.
9. The only fully reliable way of performing such analysis is by conducting actual closed
container tests and open fire tests on representative samples of the actual cells to be
installed. The closed container tests are needed to measure quantities of various Flammable
Gases; the open fire tests are need to measure quantities of the various Acute Toxic gases
such as hydrogen fluoride and carbon monoxide that are generated in various conditions of
air supply or fire suppression. Explosive behaviour is determined for regulatory purposes by
tests closely specified in the UN Manual of Tests and Criteria (having legal force under the
CLP Regulation).
10. In the absence of such explicit tests, this paper takes reports and measurements from the
technical and scientific literature to make estimates of quantities of Hazardous Substances
listed in Parts 1 and 2 that may credibly be generated, in accidents, from given quantities of
“Substances S” (i.e. battery chemicals) that are to be regarded themselves as Hazardous
Substances under Part 3.
11. The estimates given are from the known technical literature, and obviously subject to
variation depending on actual details. They represent prima facie indications of likely
behaviour. Uncertainty can only be resolved by actual measurements, per item 9.
12. Because Li-ion battery chemistry is, in detail, a closely guarded trade secret, explicit
quantities of multiple chemical components are rarely declared. However most of the
literature reports the energy storage capacity of the battery cells, and this will be essentially
proportional to chemicals content, by multiplication of identical cells, which may be many
thousands in grid-scale BESS. Hence energy storage capacity in MWh is used as the
determinant of whether Controlled Quantities (CQs) of Substances S have been exceeded.
This approach also avoids needing to list separately Controlled Quantities (CQs) of individual
battery component chemicals (Substance(s) “S”) under Part 3.
13. “Mixed oxide” cells (e.g. LCO, NMC and other acronyms describing their complex chemistry)
behave qualitatively differently from LFP (lithium iron phosphate) cells, which fail less
aggressively and generate lower temperatures in accidents. However thermal runaway can
still occur and there are documented explosions and fires at LFP BESS installations. LFP cells
also generate smaller quantities of Flammables in failures. However they are also
documented to generate larger quantities of toxic gases. This paper therefore contrasts
“mixed oxide” and LFP cells where data is available.
6
14. Physical Hazards include generation of Flammable Gases under anoxic conditions which
have, when mixed with air, led to several well-documented explosions. From the estimates
in this paper, the Controlled Quantities (CQs) are exceeded for an installation of 23.5 MWh
for the “mixed oxide” cell types. The same cell type exceeds the Controlled Quantity for
carbon monoxide alone at a capacity of 45.7 MWh. For cells qualifying as P1a Explosives, the
Controlled Quantities are exceeded for installations of around 1.55 MWh capacity.
15. Health Hazards include the generation of highly toxic hydrogen fluoride and other toxic
gases (including carbon monoxide, which has a dual classification). From the estimates in
this paper, the Controlled Quantities (CQs) are exceeded for the LFP cell types at a capacity
of 16.7 – 22.1 MWh, depending on assumptions regarding carbon monoxide. The Controlled
Quantity of hydrogen fluoride alone is exceeded at 25 MWh capacity, given LFP cells.
16. Environmental Hazards could include the generation of CQs of contaminated fire water if
used. The Controlled Quantity is readily exceeded by the volumes of water conceivable. A
contaminant potentially applicable to all cell types is copper oxide generated by burning
copper foil used in the cells. Copper oxide is listed as Acute Toxic to the aquatic environment
at low concentrations. Other highly toxic contaminants such a cobalt compounds may be
generated from some of the “mixed oxide” cell types. Data on generation of contaminants
generated in fires is however lacking, so the environmental hazard requires further
investigation, but important because complete oxidation of the copper foil in fire could
imply a particularly low threshold for HSC, as low as 1 MWh, for any Li-ion cell type.
17. For Nickel-based cathodes (e.g. NMC), the generation of inhalable nickel oxides in the form
of smoke is reported in fire tests and supported by experts. The very low Controlled Quantity
for such inhalable dusts would require HSC at a threshold around 2.75 MWh storage.
18. The various thresholds estimated in this paper below 50 MWh of energy storage capacity
are summarised in Table 13, for various stated cell types and conditions. Any one of these
being satisfied would imply, prima facie, a need to seek HSC. Per item 9, the only fully
reliable way to determine a requirement would be actual fire tests on representative
samples of the cells. It appears to be the responsibility of the operator to determine, and
notify, the requirements; this responsibility is explicit under the related COMAH Regulations.
19. Referring to Table 13, a BESS of 50 MWh capacity based on LFP cells would exceed the
Controlled Quantity CQ for the highly toxic hydrogen fluoride gas more than two-fold,
making no assumptions regarding the amount of carbon monoxide generated in fires. From
the literature cited, it is unlikely that such installations would escape an HSC obligation,
unless actual fire tests show objectively that Toxic Gas generation is below the cited
estimates. If the cells behave explosively under the specified tests for explosives, then the
obligations would commence at around 1.55 MWh (Division 1.3) or 7.75 MWh (Division 1.4).
20. A BESS of 50 MWh based on mixed oxide cathodes (e.g. NMC, LCO etc) would exceed the
Controlled quantity of “P2 Flammable gases” at 28.3 MWh. It would exceed the CQ of
carbon monoxide alone (as a Flammable Gas) at 45.7 MWh. As for LFP cells, from the
literature sources cited, it is unlikely that such installations would escape HSC obligations.
Mixed Oxide BESS are less likely to exceed CQs of Toxics, but more likely to exceed CQs of
Flammables.
21. Operation of Li-ion BESS at 50 MWh storage, of any electrode type, without HSC, is thus
likely to constitute an offence under S. 23 of the Planning (Hazardous Substances) Act 1990.
7
1.0 The Planning (Hazardous Substances) Act 1990
The designation (in England) of the District Council as the Hazardous Substances Authority (HSA) for
installations sited in their area derives from the Planning (Hazardous Substances) Act 1990
1
( P(HS)A
1990 ) s.1. Various special cases are given, but the default HSA is the District Council.
The Act controls the presence (actual, planned, or foreseeable) of “hazardous substances” (HS)
at industrial sites ( “in, over or under the land”
2
s.4(1) ) as part of the Planning process. Where
involved in any site at or above “controlled quantities” (s. 4(2)) the applicant must make application
for Hazardous Substances Consent (HSC) to the HSA which may grant or refuse the application, or
grant subject to conditions.
It is an offence under s.23 of the Act
3
for hazardous substances present on any site without HSC
being applied for and granted. The persons liable for that offence include:
s. 23(3)(b) the person in control of the land i.e. the operator;
s. 23(3)(a)(i) any person knowingly causing HS be present;
s. 23(3)(a)(ii) any person allowing HS to be so present.
Fines imposed by the Court may be scaled according the financial benefit derived from the
offence, s. 23(4A).
The HSA has the power under s. 24(1)
4
to serve a “hazardous substances contravention notice”
which may specify such steps as appear to be required to remedy the contravention.
The meaning of “any person allowing HS to be present” in s. 23(3)(a)(ii) is not clear, but both
Councillors and officers may need to consider if they are unlawfully “allowing HS to be present” by
failing to require applications for HSC in respect of BESS above a certain size.
What constitutes a “hazardous substance” or the “controlled quantity” is not specified in the
Act. These are determined by Regulations made by the Secretary of State under s. 5(1)(a)
5
.
The present Regulations are the Planning (Hazardous Substances) Regulations 2015, (S.I. 627 of
2015)
6
which provide detailed definitions of hazardous substances, controlled quantities and
procedures to be followed. These are commonly referred to as “the HSC Regulations” although that
is not their formal title, and the legislation requiring HSC is the P(HS)A 1990, not the Regulations
themselves.
1
https://www.legislation.gov.uk/ukpga/1990/10/section/1
2
https://www.legislation.gov.uk/ukpga/1990/10/section/4
3
https://www.legislation.gov.uk/ukpga/1990/10/section/23
4
https://www.legislation.gov.uk/ukpga/1990/10/section/24
5
https://www.legislation.gov.uk/ukpga/1990/10/section/5
6
https://www.legislation.gov.uk/uksi/2015/627/introduction/made
8
1.1 The Planning (Hazardous Substances) Regulations 2015 (“the HSC Regulations”)
The current Regulations were made in 2015, under the authority of the P(HS)A 1990 and also of
the European Communities Act 1972. In their detailed form they derive from the requirements of
the “Seveso III Directive” 2012/18/EU
7
and implement the land use planning aspects of the Seveso
Directive.
The HSC Regulations specify procedural matters, such as the requirement to publish notices of
applications under Reg. 6(1)
8
, the forms to be supplied, the notification of applications to the
“COMAH competent authority” under Reg. 9(1)(b)
9
etc. However the most important aspect is that
they define “for the purpose of the Act” (Reg. 3)
10
what are “hazardous substances” (HS) (Reg. 3(a))
and what are the “controlled quantities” (CQ) (Reg. 3(b)).
It is noteworthy that Reg. 3 describes “hazardous substances” as being “substances, mixtures or
preparations, present as raw materials, products, by-products, residues or intermediates” – which
covers a very wide range of forms in which “substances” may be present, to be covered by the Act
(P(HS)A 1990).
The hazardous substances HS and their controlled quantities CQ (always listed by weight, in
tonnes) are itemised in Schedule 1 to the Regulations
11
, which has three Parts, each in two
Columns. Column 1 lists the hazardous substances HS (Reg. 3(a)) and Column 2 lists the controlled
quantities CQ in tonnes (Reg. 3(b)).
Part 1 (“Categories of Substances”) lists hazardous substances according to the “hazard
categories” contained in the CLP (Classification Labelling and Packaging) Regulations. These include
Health Hazards (Section H), Physical Hazards (Section P), Environmental Hazards (section E) and
“Other” Hazards (Section O). To decide if a particular substance falls within any of these hazard
categories require consulting the CLP Regulations and the very many listings of chemicals contained
therein, and their defined hazard categories. Reg. 3(a)(i) specifies that any substance falling into
one of these categories in Part 1 is a “hazardous substance” for the purposes of the Act.
Part 2 (“Named Hazardous Substances”) lists named hazardous substances which may or may
not have a CAS
12
number [ “only for indication” ], and specifies particular substances with well-
known associated hazards and specific controlled quantities which may be different from the CQ in
Part 1 determined on the basis of the Hazard Category of the substance. Reg 3(a)(ii) specifies that
any substance specified in Part 2 is a hazardous substance for the purposes of the Act.
Definition for this paper: For clarity in what follows, in this paper we will refer to a “listed
hazardous substance” or “listed HS” to be any HS specified in either Part 1 or Part 2. This is to
distinguish these HS clearly from the hazardous substances description given in Part 3.
7
https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A32012L0018
8
https://www.legislation.gov.uk/uksi/2015/627/regulation/6/made
9
https://www.legislation.gov.uk/uksi/2015/627/regulation/9/made
10
https://www.legislation.gov.uk/uksi/2015/627/regulation/3/made
11
https://www.legislation.gov.uk/uksi/2015/627/schedule/1/made
12
CAS=Chemical Abstracts Service, a division of the American Chemical Society, a US-based chemicals database. A CAS
Registry Number or “CAS number” is a unique serial number identifying a particular chemical compound.
9
Part 3 (“Substances used in Processes”) defines a hazardous substance as:
Where it is reasonable to foresee that a substance falling within Part
1 or Part 2 (“HS”) may be generated during loss of control of the
processes, including storage activities in any installation within an
establishment, any substance which is used in that process (“S”).
Reg. 3(a)(iii) defines a HS as one “meeting the description in Column 1 of Part 3 of that
Schedule” i.e. Schedule 1 to the Regulations. “The description in Column 1” is quoted above.
In other words, any substance “S” whatever is to be considered a hazardous substance (HS) for
the purposes of the Act, under Reg 3(a)(iii) and Schedule 1 Part 3 Column 1 — provided only that it
is “reasonable to foresee” the listed HS (in Parts 1 or 2) being “generated [from S] during loss of
control of the processes”.
The Controlled Quantity (CQ) of the substance(s) “S” (e.g. the chemical components of battery
cells) is defined in Column 2 of Part 3, and reads:
The amount of S which it is believed may generate (on its own or in
combination with other substances used in the relevant process) an
amount equal to or exceeding the controlled quantity of the HS in
question.
In other words, the CQ of substances “S” (e.g. battery chemical components) are those amounts
which may generate quantities of listed HS (i.e. Hazardous Substances listed in Parts 1 or 2) above
the CQs for those listed HS, “during loss of control of the processes”.
The provisions of Part 3 are frequently overlooked, consistently so in Planning Applications for
BESS installations across the UK, although the point has been raised in respect of individual
Planning Applications, and one example will be cited below. Yet the Regulations are the UK
implementation of the Seveso III Directive, and have their origin in the Seveso disaster of 1976,
which was itself a “loss of control” accident. The entire history of the Seveso directive shows that it
was always intended to cover not only hazardous substances present by design, but also hazardous
substances generated in accidents, where “control of the processes is lost”.
Moreover, “loss of control” accidents in Li-ion BESS are the principal safety hazard that these
systems present to the public and to the environment. In normal operation, they will be innocuous,
and the chemicals on which the battery technology is based are either not “listed HS” at all, or else
are typically present below the CQs. In other words, Parts 1 and 2 are unlikely to compel the
operator of a Li-ion BESS to seek HSC.
Part 3 is an entirely different matter. The well-known phenomenon of “thermal runaway” is a
known failure mode of Li-ion batteries, and is equally well-known to generate HS in the hazard
classes H1 and P2 of Part 1 and at least one “Named HS” in Part 2 when such accidents occur. The
substances “S” are then all the chemicals integral to the Li-ion cells, which under Part 3 and Reg.
3(a)(iii) are to be considered as “hazardous substances” (HS) for the purposes of the Act, even
though not listed in Parts 1 or 2, because they have the potential to generate “listed HS” during
“loss of control of the processes” i.e. if thermal runaway or a “battery fire” occurs.
10
1.2 Hazardous Substances and Controlled Quantities under Part 3:
Part 4 of Schedule 1
13
contains various Notes to Parts 1 to 3. Note 4 in Schedule 1 says:
The controlled quantities set out in Parts 1 to 3 of this Schedule relate to each
establishment.
The quantities to be considered for the application of these Regulations are the maximum
quantities which are present or are likely to be present at any one time.
The first sentence shows that the Controlled Quantities are the aggregate quantities contained
in each “establishment” i.e. land under the control of a single operator. It is the mere presence of
the HS above the CQ which requires HSC; spatial arrangement of plant or containers is irrelevant,
and is mentioned nowhere in the Act or in the HSC Regulations.
In the second sentence, with regard to Part 3, it must be remembered that the regulated
“hazardous substances” under Part 3 are any and all of those substances “S” that may generate
listed HS in loss of control accidents. It is those “substances “S” that are the regulated “hazardous
substances” under Part 3, not the “listed HS” that may be generated in loss of control. Therefore,
the “maximum quantities which are present or likely to be present at any one time” can only refer
to the maximum quantities of such substances “S”. In the case of BESS, this is the maximum
aggregate chemical components of all the battery cells present in the “establishment”.
It would be inconsistent to read Note 4 as referring to the “maximum quantities of listed HS that
are likely to be generated during particular loss of control accidents.” This is simply not what this
Note to the Regulations says. To interpret Note 4 in this way would be wholly inconsistent with the
application of the HSC Regulations to a substance listed in Parts 1 or 2.
It should also be noted that nothing in the Regulations requires consideration of the likely
extent of a “loss of control” situation, or the likelihood of a major accident relative to a localised
incident. The Regulations require only that it be “reasonable to foresee” that a listed HS may be
generated in loss of control accidents.
The controlled quantities of substances “S” under Part 3 are then clearly defined: the CQ(s) are
those quantities of substance(s) “S” required to generate an amount exceeding the CQ of any listed
HS in Parts 1 or 2, taking account of the Aggregation Rule given in the following Note 5.
13
https://www.legislation.gov.uk/uksi/2015/627/schedule/1/made
11
1.3 The “Aggregation Rule” of Note 5:
It is critical also to consider Note 5, sometimes called
14
the “Aggregation Rule”, which specifies
how multiple HS in the same hazard class are to be aggregated to determine if the CQ has been
exceeded. If no single substance is present (or “foreseeably generated”) above its CQ, then the
Regulations still apply if quantities of multiple HS are aggregated according to the Rule given in
Note 5:
The following rule governing the addition of hazardous substances, or categories of
hazardous substances, applies where appropriate.
In the case of an establishment where no individual hazardous substance is present in a
quantity above or equal to the relevant controlled quantity, the following rule must be applied
to determine whether the establishment is covered by the relevant requirements of these
Regulations.
These Regulations apply to establishments if the sum
q1/QL1 + q2/QL2 + q3/QL3 + q4/QL4 + q5/QL5 + … is greater than or equal to 1,
where
qx = the quantity of hazardous substance x (or category of hazardous substances) falling
within Part 1 or Part 2 of this Schedule; and
QLX = the relevant controlled quantity for hazardous substance x (or category of hazardous
substances x ) from Column 2 of Part 1 or from Column 2 of Part 2 of this Schedule (except
for those substances for which column 2 contains a quantity Q*, in which case, for Hydrogen,
Q is equal to 5, and for Natural Gas (including liquefied natural gas), Q is equal to 50).
15
This rule must be used to assess the health hazards, physical hazards and environmental
hazards. It must therefore be applied three times—
(a) for the addition of hazardous substances listed in Part 2 that fall within acute toxicity
category 1, 2 or 3 (inhalation route) or STOT SE category 1, together with hazardous
substances falling within section H, entries H1 to H3 of Part 1;
(b) for the addition of hazardous substances listed in Part 2 that are explosives, flammable
gases, flammable aerosols, oxidising gases, flammable liquids, self-reactive substances
and mixtures, organic peroxides, pyrophoric liquids and solids, oxidising liquids and solids,
together with hazardous substances falling within section P, entries P1 to P8 of Part 1;
(c) for the addition of hazardous substances listed in Part 2 that fall within hazardous to the
aquatic environment acute category 1, chronic category 1 or chronic category 2, together
with hazardous substances falling within section E, entries E1 and E2 of Part 1.
The relevant provisions of these Regulations apply where any of the sums
obtained by (a), (b) or (c) is greater than or equal to 1.
14
The COMAH Regulations 2015: Guidance on Regulations L111 (3rd edition) ISBN 978 0 7176 6605 8
https://www.hse.gov.uk/pubns/priced/l111.pdf See page 92 of 132 Guidance Note 384.
15
The majority of named hazardous substance in Part 2 have the same controlled quantities as for “lower tier”
establishments under the COMAH Regulations. However the CQs for Hydrogen (entry 15) and for Natural Gas (entry 18)
are lower, being 2 tonnes and 15 tonnes respectively under the HSC Regulations, and are marked with an asterisk, i.e.
“Column 2 contains a quantity Q* ”. Where the establishment does not exceed these quantities for Hydrogen or for
Natural Gas alone, this qualification under the Aggregation Rule in Note 5 reverts the applicable quantities for
aggregating Hydrogen or Natural Gas to those listed for the same Rule under the COMAH Regulations, viz. 5 tonnes and
50 tonnes respectively.
12
1.4 Other relevant Notes in Part 4:
Note 1: Substances and mixtures are classified in accordance with the CLP Regulation.
This refers primarily to the hazard categories used in Part 1. However it has been argued that
batteries are classified as “articles” under the CLP Regulation, and hence not “substances” at all as
regards application of the law. Under this argument, there can be no “substance S” to consider
within Part 3, so Part 3 would not apply.
However this is not the case. Batteries are classified under the REACH Regulation as “articles
with integral substance(s)”; hence the “substance(s) S” referred to in Part 3 are simply those
“substances integral to the article” in the meaning of the CLP Regulation i.e. the chemical
components of the battery cells. This point is examined in more detail elsewhere.
Note 2: Mixtures shall be treated in the same way as pure substances provided they remain within
concentration limits set according to their properties under the CLP Regulation, or its latest adaptation to
technical progress, unless a percentage composition or other description is specifically given.
This Note confirms in respect of mixtures the language of Reg. 3(a):
hazardous substances are substances, mixtures or preparations— [in Parts 1 2 or 3]
… and present as raw materials, products, by-products, residues or intermediates;
Note 6: In the case of hazardous substances which are not covered by the CLP Regulation, including
waste, but which nevertheless are present, or are likely to be present, in an establishment and which
possess or are likely to possess, under the conditions found at the establishment, equivalent properties
in terms of major accident potential, these must be provisionally assigned to the most analogous
category or named hazardous substance falling within the scope of these Regulations.
This Note shows that “unlisted” substances outside the scope of the CLP Regulation must
nevertheless be assigned provisionally to the most analogous category within the scope of the HSC
Regulations. Only a fraction of the registered chemicals in the ECHA database have Harmonised
Classifications; that is an ongoing “work in progress”. Whether a chemical has a Harmonised
Classification or not, it still falls under the CLP Regulation once it has any of the hazardous
properties defined by the UN GHS classification methodology. An example would be the toxic (and
unstable) gas Phosphoryl Fluoride POF3 known to be evolved from BESS “fires” but which lacks a
Harmonised Classification under the CLP Regulation.
This Note also shows that “waste” (which is not covered by the CLP Regulation) is nevertheless
covered by the HSC Regulations. “Waste” will always be ill-defined as to chemical composition and
likely to be mixed with many chemical substances, parts, objects, and could well include hazardous
substances contained within “articles”.
It is clear that the drafters did not intend there to be “loopholes” for hazardous substances, even if
such are not covered by the CLP Regulation. This Note effectively refutes the HSE position that BESS
are “articles” (not covered by CLP) and therefore exempt.
Note 7: In the case of hazardous substances with properties giving rise to more than one classification,
for the purposes of these Regulations the lowest controlled quantities apply. However, for the application
of the rule in note 5, the lowest controlled quantity for each group of categories in notes 5(a), 5(b) and
5(c) corresponding to the classification concerned must be used.
13
This Note resolves potential ambiguities in dual classifications in Part 1, and with application of the
Aggregation Rule Note 5. For example, Carbon Monoxide is both toxic (section H) and inflammable
(section P) and therefore has a dual classification under the sections of Part 1.
Note 9: The hazard class Explosives includes explosive articles (see Section 2.1 of Annex I to the CLP
Regulation). If the quantity of the explosive substance or mixture contained in the article is known, that
quantity must be considered for the purposes of these Regulations. If the quantity of the explosive
substance or mixture contained in the article is not known, then, for the purposes of these Regulations,
the whole article must be treated as explosive.
Note 10: Testing for explosive properties of substances and mixtures is only necessary if the screening
procedure according to Appendix 6, Part 3 of the UN Recommendations on the Transport of Dangerous
Goods, Manual of Tests and Criteria (UN Manual of Tests and Criteria)(1) identifies the substance or
mixture as potentially having explosive properties.
Notes 9 and 10 raise the possibility that BESS cells should be considered as “explosive articles”.
BESS cells can certainly fail explosively, but this behaviour depends of the cell’s State of Charge
(SoC). The point will be examined further below.
Note 23: Where a hazardous substance falls within both Parts 1 and 2 of this Schedule, the controlled
quantity in Part 2 applies.
This Note resolves ambiguities between substances classified in both Parts 1 and 2.
Note 24: In relation to Part 3—
(a) where S also falls within Part 1 or Part 2, the classification with the lowest controlled quantity applies;
and
(b) where S also falls within Part 1 and Part 2, the controlled quantity which is lowest when the controlled
quantities under Part 2 and Part 3 are compared applies.
This Note resolves ambiguities between substances S controlled under both Part 3, and Parts 1 or 2.
14
2.0 Applicability of the HSC Regulations to BESS
In relation to Li-ion BESS, not only is it “reasonable to foresee” that listed HS may generated “if
control of the processes is lost”, it is certain, and well-documented in the technical literature.
Moreover, the failure mode of “thermal runaway” is well-known and an increasing list of BESS
“fires” in many parts of the world shows that such accidents are now matters of record, not
speculation. It is now impossible to argue that generation of listed HS in loss of control accidents
would not be “reasonable to foresee”: such generation, and the occurrence of loss of control
accidents in BESS, are documented facts.
Therefore, under Part 3, any substance “S” used in the BESS cells should be considered a
Hazardous Substance, and it is those cell component chemicals “S” that are controlled, and have
the controlled quantities CQ applied to them.
The CQs are not the quantities of listed HS predicted to be generated in any particular accident.
Nothing in the Regulations depends on likelihood of occurrence, or predicted extent of accidents;
the Regulations simply regulate the total inventory of hazardous substances.
Under the loss of control considerations of Part3, any substances “S” “foreseeably generating”
listed HS are to be considered hazardous substances themselves, and the CQ is that amount of “S”
which may generate a CQ of a listed HS in loss of control.
It is hard to see how grid-scale BESS can escape the “loss of control” provisions of Reg 3(a)(iii)
and Part 3 of Schedule 1.
It is equally hard to see how the quantity determining the requirement for HSC under P(HS)A
s.4(2)
16
can be other than the total inventory of battery chemical components “S”, regulated as
hazardous under Part 3, that are present in the “establishment” (P(HS)A s.4(2)(aa)).
The technical question requiring consideration is therefore to determine the CQ of those
substances “S”, being integral components of the battery cells, which would generate controlled
quantities of listed HS “if control of the process is lost”, aggregated over the whole establishment.
Some estimates are made in subsequent sections.
2.1 Determination of CQ of integral substance(s) S considered hazardous under Part 3
From the foregoing considerations, the determination of a CQ under Part 3, for BESS integral
substances “S”, requires consideration of:
i. The listed hazardous substances foreseeably generated in loss of control of the processes;
ii. the quantity of all listed HS generated from given quantities of battery components “S”;
iii. the application of the Aggregation Rule to determine if the CQs are exceeded;
iv. the separate application of the Aggregation Rule for Toxics (H), Flammables (P) and potentially
Environmental (E) hazards.
The requirement for HSC commences at the lowest BESS inventory which first results in either
an individual substance exceeding its CQ, or one particular hazard class (H, P or E) exceeding unity
when calculating the sum given in the Aggregation Rule.
16
https://www.legislation.gov.uk/ukpga/1990/10/section/4
15
2.2 Relationship of the HSC Regulations to the COMAH Regulations
Operational health and safety aspects of the Seveso Directive were implemented in the form of
the Control of Major Accident Hazards (COMAH) Regulations 2015
17
under the provisions of other
parent legislation (in particular the Health and Safety at Work etc Act 1974
18
). The COMAH
Regulations are not the concern of the local Planning Authority, however the same list of dangerous
substances together with controlled quantities appears in the COMAH Regulations, and the HSC
Regulation 9(1)(b)
19
requires the HSA to notify the “COMAH competent authority” on receipt of a
HSC application. In England, the “COMAH competent authority” is the Health and Safety Executive
(HSE) plus the Environment Agency (EA) acting jointly (Reg. 4)
20
.
Apart from the change of language from “hazardous substances” (HSC Regulations e.g. Reg. 3
21
)
to “dangerous substances” (COMAH Regulations e.g. Reg. 2
22
), inspection of the Schedules to both
sets of Regulations show that the substances in question are essentially identical, being based on
the same hazard classes in the CLP (Classification Labelling and Packaging) Regulation
23
. It was
necessary for both sets of regulations to be aligned in this way in order for the UK to be compliant
with the Seveso III Directive
24
2012/18/EU; that is the reason that the Schedules prescribe
essentially the same substances.
The Controlled Quantities in the HSC Regulations correspond to the Controlled Quantities in the
“lower tier” of the COMAH Regulations. “Upper tier” in the COMAH Regulations prescribes more
stringent controls in accident prevention but those higher thresholds do not affect the operation of
the HSC Regulations.
Two named substances, viz. Hydrogen and Natural Gas, have lower Controlled Quantities for the
purposes of the HSC Regulations than they do for the purposes of the COMAH Regulations, but
these are specific exceptions.
17
https://www.legislation.gov.uk/uksi/2015/483/introduction
18
https://www.legislation.gov.uk/ukpga/1974/37/contents
19
https://www.legislation.gov.uk/uksi/2015/627/regulation/9/made
20
https://www.legislation.gov.uk/uksi/2015/483/regulation/4
21
https://www.legislation.gov.uk/uksi/2015/627/regulation/3/made
22
https://www.legislation.gov.uk/uksi/2015/483/regulation/2
23
https://www.legislation.gov.uk/eur/2008/1272/contents ; also https://eur-lex.europa.eu/legal-
content/EN/TXT/PDF/?uri=CELEX:32008R1272 and “Guidance in the application of the CLP criteria” Regulation EC
1272/2008 on classification labelling and packaging of substances and mixtures” ECHA version 5.0 July 2017
https://echa.europa.eu/documents/10162/23036412/clp_en.pdf/58b5dc6d-ac2a-4910-9702-e9e1f5051cc5
24
https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A32012L0018
16
3.0 Hazardous Substances under Parts 1 or 2 generated in loss of control accidents in BESS
The principal failure mode (“loss of control of the processes”) that has to be considered is that
of “thermal runaway” described in a previous report
25
. Popularly called a “battery fire” this need
not involve flames at all, but generation of a toxic and inflammable “vapour cloud” that if mixed
with air and ignited may cause a rapid deflagration or “vapour cloud explosion”. The 2019 accident
at McMicken, Arizona
26
, was of this nature, and did not involve flames except in the explosion
itself
27
. It derived just one rack out of 27 in a relatively small BESS, but resulted in a dramatic
explosion that caused life-limiting injuries to three out of four first-responders.
Actual fires are also possible and have occurred in the increasing catalogue of major BESS fires,
and even in a fully discharged BESS would probably involve combustion of the organic components
of the cells, producing gases “typical of a plastics fire”.
The only fully reliable method of determining a complete list of hazardous substances that
might be generated in loss of control situations would be to perform actual “fire tests” with
comprehensive chemical analysis, including fire-fighting situations involving water cooling, and
contaminated “fire water”. Such tests should ideally be done on representative samples of the
actual battery cells proposed to be installed, whose exact composition is usually a strongly
protected trade secret. Such comprehensive tests have rarely if ever been performed, but a list of
leading contenders may be derived from reports in the technical literature, both from laboratory
reports and published scientific papers, and engineering failure analyses of major BESS “fires” that
have received post hoc investigation.
Flammable and Toxic Gases: From the technical literature, the gases listed in Table 1 are
potentially generated in loss of control accidents.
Some of these, such as Hydrogen Cyanide (HCN) and Hydrogen Chloride (HCl), are probably
generated from incidental packaging or insulation polymers such as polyurethane (PU) or poly(vinyl
chloride) (PVC); Li-ion BESS chemicals are dominated by fluorine-containing polymers, and are
predominantly not nitrogen-containing nor chlorine-containing. Nevertheless, HCN in particular is a
well-known hazard to fire-fighters, a common cause of death in fires, and was an explicit cause of
concern to professional fire-fighters in the 2019 McMicken incident. Also, expert reports indicate
that “a common signature of a suspected thermal runaway event is the presence of HCN, HCl, or
HF”, so HCN and HCl toxic gases require consideration from an operational and fire-fighting
perspective even if they may be minor components.
The remaining gases fall into two groups. Hydrogen, Ethylene, Methane and Carbon Monoxide
are all known to be evolved in thermal runaway and are classified as “P2 Flammable Gases” under
the HSC Regulations. Hydrogen Fluoride and Phosphoryl Fluoride are evolved in thermal runaway,
25
Fordham E J, Allison W & Melville D (2021) “Safety of Grid-scale Li-ion Battery Energy Storage Systems”, 5 June 2021
https://www.researchgate.net/publication/352158070_Safety_of_Grid_Scale_Lithiumion_Battery_Energy_Sto
rage_Systems
26
Hill D (2020). McMicken BESS event: Technical Analysis and Recommendations, Arizona Public Service.
Technical support for APS related to McMicken thermal runaway and explosion.” Report by DNV-GL to Arizona Public
Service, 18 July 2020. Document 10209302-HOU-R-01
27
McKinnon M B, DeCrane S, Kerber S (2020). Four fire-fighters injured in Lithium-ion Battery Energy Storage System
explosion – Arizona. Underwriters Laboratories report July 28, 2020. UL Firefighter Safety Research Institute, Columbia,
MD 20145
17
and in subsequent air-dependent fires Carbon Monoxide will continue to be produced where
combustion of organic materials or graphite from battery anodes is incomplete. These gases are
classified as “H1 or H2 Acute Toxic” health hazards (CO is both combustible and toxic).
Environmental hazards (substances Toxic to the aquatic environment): Finally, the possibility of
substances Toxic to the Aquatic Environment must be considered. The most serious source of
aquatic toxic substances in a loss of control event at a BESS arises when water is used in fire-
fighting. Although the Controlled Quantity of substances E1 “Hazardous to the Aquatic Environment
in Category Acute 1 or Chronic 1” is 100 tonnes, this tonnage of contaminated fire water easily
arises if a cumulative amount of water exceeding 100 m3 is deployed in firefighting.
Moreover, the CQ of E1 Environmentally hazardous substances need not refer to a tonnage of
pure substance; generally speaking they do not. The tonnage refers to the contaminated fire water
(or other similar mixture). The concentrations of toxic materials which may be regarded as “toxic to
the aquatic environment” may be quite low, and are specified by means “M-factors” (short for
“multiplying factor”)
28
which are assigned for substances classified as toxic to the aquatic
environment on the basis of LC50 or EC50 or NOEC concentrations in tests against fish, crustacea or
plants. Details are explained in Annexes to the CLP Regulation.
In the practical application of M-factors, what matters for the purposes of this paper is the
concentration at which contaminated fire water would be considered as “E1 Toxic to the aquatic
environment”. This is itself likely to be a complex mixture in fire situations, and toxic substance
concentrations must be summed according to the “Summation Method”
29
. This is a technical part
of the CLP Regulation and a different summation from that required in the Aggregation Rule of
Note 5 to the HSC Regulations; the Summation Method is a method for classifying mixtures of
environmentally toxic substances according to the environmental toxicity assessed for the
substances.
For substances classified as “Acute 1”, the Summation method is a simple addition (more
complicated sums are defined for mixtures of substances with lower Categories of toxicity). Then
30
a mixture is classified as “Acute 1” if the sum of concentrations of toxic components multiplied by
their M-factors is ≥ 25%. Similar considerations
31
apply for mixtures to be classified as “Chronic 1”.
This means for example that a 2.5% concentration (by weight) of a substance classified as Acute
Toxic 1 with a M-factor of 10 is to be classified as Acute Toxic (to the aquatic environment)
Category 1, and similarly for Chronic Toxicity. Both Category 1 classifications are classified as E1
under the HSC Regulations. Chronic Category 2 mixtures are classified E2 under the HSC
Regulations.
Cobalt (II) oxide is an oxide conceivably generated in oxidising fires from cells with cobalt-
containing cathodes. An M-factor of 10 is listed; if present at 2.5% by weight in water, this would be
classified as an E1 environmental hazard. If additional toxic metal oxides or other compounds with
28
See, e.g., https://www.chemsafetypro.com/Topics/GHS/What_is_M-
factor_and_how_it_is_used_for_enviromental_hazard_classification.html
29
Annex I to the CLP Regulation, section 4.1.3.5.5. See Guidance document
https://echa.europa.eu/documents/10162/23036412/clp_en.pdf/58b5dc6d-ac2a-4910-9702-e9e1f5051cc5
30
Rule defined in Table 4.1.1 of Annex I to the CLP Regulation, also in the above Guidance document.
31
Table 4.1.2 of Annex I to the CLP Regulation.
18
M-factor 10 were present, the contaminated fire water becomes a E1 Environmental hazard if the
summation of all toxic compounds exceeds 2.5% by weight in water.
Some potentially relevant environmentally toxic substances are listed in Table 2 with M-factors
where available. One (cobalt lithium nickel oxide) is an actual Li-ion battery cathode material. The
others are toxic substances conceivably generated during BESS “fires” or thermal runaway events.
Some of these oxides will occur only for certain electrode chemistries, for example the popular
“NMC” (nickel manganese cobalt” cathode chemistry. The “LFP” (Lithium iron phosphate) cathode
chemistry does not contain cobalt or nickel so the cobalt compounds would not be present.
However all classes of Li-ion battery have the same anode concept, which is finely divided
graphite (carbon) supported on a thin copper foil. Large tonnages of copper foil will be present in
grid-scale BESS which are likely to oxidise in fires to form copper oxides; both copper (I) oxide and
copper (II) oxide are Acute and Chronic Toxic to the aquatic environment in Category 1. Moreover
both these oxides of copper have a large M-factor (100) for acute toxicity in the aquatic
environment, meaning that solutions or dispersions (mixtures) of these oxides in water as low as
0.25% by weight are classified as Acute Toxic, Category 1, and E1 environmental hazards under the
HSC Regulations.
Therefore, fire water contaminated with copper oxides at only 0.25% is classified as an E1
environmental hazard under the HSC Regulations. A fire-fighting operation using 100 m3 of fire
water then exceeds the Controlled Quantity of E1 Environmental Hazard, provided only that the fire
water is contaminated with 0.25% by weight (i.e. 0.25 tonnes or 250 kg) of copper oxides generated
in a large fire.
Other copper compounds potentially generated are copper fluoride CuF2, which forms in
oxidising conditions from Cu metal, oxygen and HF at temperatures above 400 °C, with water
(steam) as the other reaction product
32
. This is a conceivable reaction in a fire also generating HF
gas; the CuF2 is slightly soluble in water and decomposes in hot water forming copper hydroxide
33
,
another listed compound with a harmonised Classification as Acute and Chronic Toxic, Category 1.
Transition (Group III) metals such as Cobalt, Nickel, Manganese and Copper are well-known to
form compounds which are persistent and toxic to living systems except in trace quantities. It is no
surprise therefore that listed compounds already have Aquatic Toxicity classifications and assigned
M-factors for application to the CLP Regulation and hence to the HSC Regulations. Whilst some of
the problems associated with the NMC electrode chemistry could be avoided by using the LFP
chemistry, because all designs of Li-ion battery to date use copper foil as a support for graphite in
the anodes, the environmental toxicity problems associated with copper compounds cannot be
avoided.
The variety of cathode chemistries in Li-ion battery cells, the ubiquitous presence of copper foil
as an anode structure and the toxic compounds that may be generated in BESS fires all warrant
further investigation for the application of the Environmental Hazard section of the HSC Regulations
(and of course for environmental protection in general).
32
Subramanian, M.A. and Manzer, L.E. (2002) Science 297 1665 https://www.science.org/doi/10.1126/science.1076397
33
Wikipedia entry for Copper (II) fluoride
19
Name of
Gas
Chemical
formula
HSC (“Seveso”) hazard category
GHS34 Hazard Codes, Classes and Categories
as listed in the C&L Inventory of ECHA or GB
Mandatory Classification and Labelling List
Controlled
Quantity
(tonnes)
Source of data
Hydrogen
H2
P2 Flammable Gas (Part 1)
Named substance (Part 2)
H220 Flammable Gas, Cat. 1A
2 (named, HSC)
5 (COMAH) & for
Aggregation Rule
HSC Regs
C&L Inventory35
Ethylene
C2H4
P2 Flammable Gas
H220 Flammable Gas, Cat. 1A
H336 STOT SE, Cat. 3
10 (P2)
HSC Regs
C&L Inventory
Methane
CH4
P2 Flammable Gas
H220 Flammable Gas, Cat. 1A
10 (P2)
HSC Regs
C&L Inventory
Carbon
Monoxide
CO
P2 Flammable Gas
H2 Acute Toxic, Cat. 3, inhalation
H220 Flammable Gas, Cat 1A.
H331 Acute Toxicity, inhalation, Cat. 3
H372 STOT RE 1
H360D Reproductive Toxicity, Cat. 1
10 (P2)
50 (H2)
HSC Regs
C&L Inventory
Hydrogen
Chloride
HCl
H2 Acute Toxic, Cat. 3, inhalation
Named substance (Part 2) if liquefied
H331 Acute Toxicity, inhalation, Cat. 3
H314 Skin corrosion, Cat. 1A,B,C
50 (H2)
25 (named)
HSC Regs
C&L Inventory
Hydrogen
Cyanide
HCN
H1 Acute Toxic (as mixture)
H2 Acute Toxic (as pure substance)
E1 Aquatic Hazard Acute, Cat. 1
E1 Aquatic Hazard Chronic, Cat. 1
H300 Acute Toxic, oral, Cat. 2
H330 Acute Toxic, inhalation, Cat. 2
H310 Acute Toxic, dermal, Cat. 1
H400 Aquatic Hazard, Acute, Cat. 1
H410 Aquatic hazard, Chronic, Cat 1
5 (H1)
50 (H2)
100 (E1)
GB Mandatory
Classification and
Labelling List,
HSE36
Hydrogen
Fluoride
HF
H1 Acute Toxic (dermal)
H2 Acute Toxic (oral, inhalation)
H300 Acute Toxic, oral, Cat. 2
H310 Acute Toxic, dermal, Cat. 1
H330 Acute Toxic, inhalation, Cat. 2
H314 Skin Corrosion, Cat. 1A
5 (H1)
50 (H2)
HSC Regs
C&L Inventory
Phosphoryl
Fluoride
POF3
Not determined but precursor of HF so
likely to be H1 Acute Toxic per Note 6
Not listed in C&L Inventory or GB MCL List but
“provisionally assigned” H310 per Note 6
5 (H1)
HSC Regs
Table 1: Gaseous Hazardous Substances generated in BESS loss of control accidents
34
GHS= Global Harmonised System is a UN-sponsored classification to which the EU and UK voluntarily adhere for the purposes of the CLP Regulation. Hazard codes are
defined and explained in multiple chemicals databases and in UNECE documents e.g. https://unece.org/DAM/trans/danger/publi/ghs/ghs_rev07/English/06e_annex3.pdf
35
The C&L Inventory is a database of the European Chemicals Agency ECHA containing many Harmonised Classifications for the purposes of the CLP Regulation
https://echa.europa.eu/information-on-chemicals/cl-inventory-
database?p_p_id=dissclinventory_WAR_dissclinventoryportlet&p_p_lifecycle=0&p_p_state=normal&p_p_mode=view
36
Harmonised Classification is only H2 for the pure substance though the great majority of Notified Classifications reckon HCN as Acute Toxic Category 1 hence H1. The HSE
GB MCL list is authoritative for GB (though not for NI) after Brexit and was therefore consulted here. “All existing EU harmonised classification and labelling in force on 31
December 2020 are retained in GB as the GB Mandatory Classification and Labelling List” https://www.hse.gov.uk/chemical-classification/legal/clp-regulation.htm
20
Table 2: Environmentally Hazardous Substances potentially generated from electrode materials in BESS loss of control accidents
(indicative only: not exhaustive)
Name of
Substance
Chemical
formula
HSC (“Seveso”) hazard category
GHS Hazard Codes, Classes and Categories
as listed in the C&L Inventory of ECHA
M-factors
Controlled
Quantity
(tonnes)
Source of
data
Cobalt (II)
Oxide
CoO
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
H370 STOT SE 1 is Notified to CLP but not
Harmonised
10
100 (E1)
EC List number
215-154-6
Cobalt (II,III)
Oxide
Co3O4
Provisionally E1 Aquatic hazard,
per Note 6
No harmonised classification found but likely to
be H400 and H410 for same reasons as for CoO
Not found
100 (E1)
EC List number
215-157-2
Cobalt Lithium
Nickel Oxide
Complex. IUPAC
name: cobalt
dihydrate lithium
hydride nickel
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H3 STOT SE Cat. 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
H372 STOT RE 1
Not found
100 (E1)
50 (H3)
EC List number
442-750-5
Copper (I)
oxide
Cu2O
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
Harmonised classification
100 (acute)
10 (chronic)
100 (E1)
EC List number
215-270-7
Copper (II)
oxide
CuO
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
Harmonised classification
100 (acute)
10 (chronic)
100 (E1)
EC List number
215-269-1
Copper (II)
hydroxide
Cu(OH)2
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
Harmonised classification
10
10 (chronic)
100 (E1)
EC List number
243-815-9
Copper (II)
Fluoride
CuF2
Provisional E1 – see text
Not found in C&L Inventory
Use value for
Cu(OH)2
100 (E1)
Listing in
Wikipedia
Dicopper
chloride
trihydroxide
Cu2Cl(OH)3
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
(Harmonised Classification)
10
10 (chronic)
100 (E1)
EC List number
215-572-9
Copper (I)
chloride
CuCl
E1 Hazard to Aquatic
Environment, Acute 1 or Chronic 1
H400 Aquatic Acute Cat. 1
H410 Aquatic Chronic Cat. 1
(Harmonised Classification)
Not found
100 (E1)
EC List number
231-842-9
21
3.1 Routes to generation of Part 1,2 Hazardous Substances in loss of control accidents in BESS:
(I) Generation of P2 Flammable gases in Anoxic conditions
An early consequence of “thermal runaway” events in BESS is known to be the generation of
inflammable gases. These are necessarily generated in anoxic conditions within the cell or the gases
would immediately burn. A cloud of combustible vapour and organic electrolyte droplets may form,
which subsequently mixes with air and may then ignite in a so-called “vapour cloud explosion” or
they may burn in other modes. The details of subsequent fire or explosion are immaterial to the
application of the HSC Regulations which are concerned solely with the quantities of the flammable
gases (or other “Physical” hazards) which may be generated in loss of control events.
One literature source, by Golubkov et al.
37
, gives quantities of gases evolved from typical Li-ion
battery cells and is widely relied upon by industry experts analysing BESS accidents. The gases
detected include the hydrocarbons in Table 1, hydrogen, and carbon monoxide and dioxide. A
limitation of the measurement system was that the known reaction product hydrogen fluoride (HF)
was not detected; this highly toxic gas can however be estimated from other sources (below).
Proportions and total quantities are given for various cell types, including the widely used
“NMC” and “LFP” electrode chemistries. A hybrid cathode, LCO-NMC, is included. Typical
compositions of whole cells (by mass) are also given, together with temperatures reached at and in
thermal runaway. The LCO-NMC and LFP types describe the range of behaviour in thermal
breakdown. An abstract of relevant parameters
38
taken from this paper is given in Table 3, including
the total amount of gases evolved. The same source provides measurements of proportions of the
different gases evolved during thermal runaway (in anoxic conditions). For application to
commercial BESS, where cell details are unlikely to be disclosed for reasons of commercial
sensitivity, but overall energy storage capacities in MWh will generally be disclosed, or required by
regulatory authorities, we need estimates of the masses of different gases evolved, per unit energy
storage in Wh (or on the large scale, in MWh). These are given in Table 4, calculated using standard
molecular weights, and the energy storage estimates in Table 3.
For application of the HSC Regulations we require quantities as masses (over the whole site, in
tonnes). For the “P2 Flammable Gases” only, the same masses are given in Table 5 scaled to a
nominal 50 MWh of energy storage, and their contributions to the summation in the Aggregation
Rule are shown in the right-most columns, for the two cell types. The energy storage at which the
Aggregation Rule sums exceed unity is also given, in MWh.
We may conclude from the data given in this published source, that BESS based on LCO-NMC
cathodes should trigger a HSC assessment at a storage capacity of 28.3 MWh, based on the
Aggregation Rule. For BESS based on LFP cells, the assessment threshold is higher, at 126 MWh. A
HSC assessment would be required for the LCO-NMC electrode type, on the basis of Carbon
Monoxide alone (CQ = 10 tonnes
39
) at 45.7 MWh.
These estimates are derived from a single source. Further accuracy could only be obtained from
actual closed container tests on representative samples of the actual cells to be installed, with
detection of hydrogen fluoride and other toxic fluorides added to the measurement system.
37
Golubkov A W, Fuchs D, Wagner J et al. (2014). Thermal runaway experiments on consumer Li-ion batteries with
metal-oxide and olivin-type cathodes. RSC Advances 4, 3633-3642 doi: 10.1039/c3ra45748f
38
The data for the plain “NMC” cathode type is omitted for clarity in this discussion. The results are intermediate
between the LOC-NMC and LFP types and qualitatively similar to the LCO-NMC.
39
As a Physical Hazard rather than a Health Hazard.
22
Cathode type:
LCO-NMC
LFP
Energy storage capacity40, Wh
9.4
3.3
Energy density41, Wh/kg
211
85
Temperature at onset of thermal runaway, °C
149 ± 2
195 ± 8
Maximum temperature reached, °C
853 ± 24
404 ± 23
Quantity of gas evolved, mmol
265 ± 44
50 ± 4
Table 3: Energy densities, thermal runaway temperatures, and total gas generation in anoxic
conditions, from two contrasting cell types in thermal runaway
Proportions42 of gases (mole %):
Masses ( mg / Wh ):
MW / g mol-1
LCO-NMC
LFP
LCO-NMC
LFP
H2
2
30.0
30.9
17.0
9.4
CO2
44
24.9
53.0
310.2
353.3
CO
28
27.6
4.8
218.8
20.4
CH4
16
8.6
4.1
39.0
9.9
C2H4
28
7.7
6.8
61.0
28.8
C2H6
30
undetected
0.3
n/a
1.36
Table 4: Published gas mole fractions, and implied masses per unit energy storage, from two
contrasting cell types in thermal runaway
Controlled
Quantity
Masses ( mg / Wh )
(equivalent to kg/ MWh)
Masses (tonne / 50 MWh)
Contribution to
Aggregation Rule
(tonnes)
LCO-NMC
LFP
LCO-NMC
LFP
LCO-NMC
LFP
H2
5 43
17.0
9.4
0.85
0.47
0.170
0.094
CO
10 44
218.8
20.4
10.94
1.02
1.094
0.102
CH4
10
39.0
9.9
1.95
0.50
0.195
0.050
C2H4
10
61.0
28.8
3.05
1.44
0.305
0.144
C2H6
10
n/a
1.36
n/a
0.068
n/a
0.0068
Aggregation Rule sums for 50 MWh Reference Case:
1.764
0.396
Storage (MWh) at which Aggregation Rule for Physical Hazards > 1 :
28.3 MWh
126 MWh
Table 5: Calculated gas masses
45
from Table 4; the same data in tonnes/50 MWh; and
contributions to the sum required in the Aggregation Rule (for a 50 MWh BESS).
Table 5 represents an application of the Aggregation Rule for the “Physical Hazards” category
only. As Note 5 makes clear, the Aggregation Rule must be applied 3 times, for the Physical, Health
and Environmental hazard classes. A subsequent section considers Health hazards.
40
Not listed directly in paper. Estimated from charge capacity in Ah and mid-range terminal voltages given in V
41
Calculated from estimated energy capacity and all-up cell weight given in paper. Typical values cited elsewhere for
these technologies c. 2014 are 200 and 90 Wh/kg respectively, so the values correspond to literature elsewhere.
42
The paper is not explicit that these are mole fractions rather than mass fractions. However they are measured by gas
chromatography which would normally be calibrated to deliver a mole fraction or mole % in the gas.
43
For the purposes of the Aggregation Rule. The CQ of H2 alone in the HSC Regulations is, exceptionally, 2 tonnes.
44
As a P2 Flammable Gas, not as a H2 Health Hazard, where a CQ of 50 tonne applies.
45
CO2 omitted because non-flammable.
23
3.2 Other Physical Hazards: P1a and P1b Explosives.
The hazard class P1a Explosives includes those substances, mixture or articles that have
Explosive properties in Divisions 1.1, 1.2, 1.3, 1.5 or 1.6 of the CLP Regulation
46
for which ECHA
Guidance Notes are available
47
.
Annex I Reg. 2.1.2.2(b) defines a Division 1.2 substance, mixture or article as:
Division 1.2: Substances, mixtures and articles which have a projection hazard but not a mass explosion
hazard;
Annex I Reg. 2.1.2.2(c) defines a Division 1.3 Explosive as:
Division 1.3: Substances, mixtures and articles which have a fire hazard and either a minor blast hazard
or a minor projection hazard or both, but not a mass explosion hazard:
(i) combustion of which gives rise to considerable radiant heat; or
(ii) which burn one after another, producing minor blast or projection effects or both;
Annex I Reg. 2.1.2.2(d) defines a Division 1.4 Explosive as:
Division 1.4: Substances, mixtures and articles which present no significant hazard:
substances, mixtures and articles which present only a small hazard in the event of ignition or initiation.
The effects are largely confined to the package and no projection of fragments of appreciable size or
range is to be expected. An external fire shall not cause virtually instantaneous explosion of almost the
entire contents of the package;
Division 1.4 Explosives do not qualify as Explosives in Hazard Class P1a but they are included in
Hazard class P1b. Hence any Explosive in Divisions 1.1 to 1.6 inclusive is a Hazardous Substance in
Part 1 of Schedule 1
48
. The CQ for P1a Explosives (Divisions other than 1.4) is 10 tonnes; the CQ for
P1b Explosives (Division 1.4 Explosives) is 50 tonne.
It is clear from Note 9 in Part 4 of the HSC Regulations that the Hazard class of Explosives
includes explosive articles. In this classification there is no question of “articles” being exempt: it is
openly explicit.
The question is thus whether Li-ion battery cells should be considered as Explosives, and if so in
which Division 1.1 through 1.6 they should be placed. The practical matter for the application of the
HSC Regulations is whether they qualify as Division 1.4 or otherwise; the CQ is 50 tonne for Division
1.4 Explosive articles (P1b Explosives) but 10 tonne for the other Divisions (P1a Explosives).
Li-ion battery cells are certainly capable of failing explosively, although this is well-known to
depend on their State of Charge (SoC). They can fail explosively either by heating (which is the
central failure risk of Li-ion cells undergoing thermal runaway) or by electrical overcharging, or by
internal failure when in a high SoC (both of the latter being typical initiating events of thermal
runaway accidents). Examples are openly available in the form of internet videos (i) for small
cylindrical consumer cells of 2.6 Ah (about 9 – 10 Wh);
49
and also (ii) what appears from the
rectangular form to be a stack of pouch cells, in a deliberate fire test initiated by overcharging.
50
46
https://www.legislation.gov.uk/eur/2008/1272/annex/I/division/2/division/2.1/division/2.1.2/division/2.1.2.2
47
“Guidance in the application of the CLP criteria” Regulation EC 1272/2008 on classification labelling and packaging of
substances and mixtures” ECHA version 5.0 July 2017
https://echa.europa.eu/documents/10162/23036412/clp_en.pdf/58b5dc6d-ac2a-4910-9702-e9e1f5051cc5
48
https://www.legislation.gov.uk/uksi/2015/627/schedule/1
49
Li-ion batteries video: 2.6 Ah consumer cells. https://www.youtube.com/watch?v=CUgbmCSmSNY
50
Li-ion batteries video: stack of pouch cells https://www.youtube.com/watch?v=EDhE0pk3FeQ
24
Frames from the first video are abstracted in Figure 1, illustrating the damage done to a non-
faulty cell, simply by overheating externally. The failure is certainly explosive, with an obvious
projectile hazard in addition to considerable heat release. The cell is fully charged.
Other examples of explosive failure with projectile hazard are shown initiated by mechanical
abuse. The same video makes the point that discharged cells do not fail explosively and are not
dangerous unless abused to the point of destruction.
Figure 1: (a) A
charged 2.6 Ah cell
being deliberately
overheated. (b) at
the point of rupture
(c) the cell takes off
as a rocket (d)
seconds later the
discharge is
complete, and the
cell is red hot. 5:56
mins to 6:24 mins.
Frames from the second video are abstracted in Figure 2, showing what appears to be a stack of
rectangular format pouch cells being tested behind a reasonably secure fire bunker. The initiating
event is deliberate overcharging; the first cell then explodes with a conspicuous fireball and
sideways dual jets of flame; later in the sequence other cells go sequentially into thermal runaway
with similar fireball behaviour as they fail.
25
Figure 2. (a) Stack of fully-
charged pouch cells in fire
bunker for thermal
runaway tests. (b) The first
cell fails explosively in a
white-hot fireball at
0:09/2:13 minutes. (c)
subsequent jets of flame
last another 8 seconds (d)
(over page) a second cell
goes into thermal runaway
(0:33/2:13 minutes) with a
similar fireball, lasting
about 5 seconds (e) (over
page) the last of 5 or 6 such
fireballs (1:55/2:13 mins).
(b): 0:09 minutes. This fireball emerges in less than 0.5 s from first precursors.
(c): The same event at 0:12 minutes, 3 seconds later
26
(d): Second cell fails at 0:33 minutes
(e): the last of 5 or 6 similar fireballs at 1:55 minutes
27
3.2.1 Field tests by Paul Christensen, Professor of Pure and Applied Electrochemistry, School of
Engineering, Newcastle University.
Figure 3 shows a sequence from a video
51
by Christensen showing the effect if mechanical damage
on a single Electric Vehicle module of stored energy 1.7 kWh, and in a 100% State of Charge. The
energy of 1.7 kWh is substantially larger than either of the previous videos, but still much smaller
than the stored energy of a grid-scale BESS, where even small storage cabins can accommodate
around 2 MWh, i.e. over 1000 times larger than the energy in the EV module shown in these videos.
Figure 3 (a). A 23 kg hammer and nail begins to fall on the EV Li-ion module:
Figure 3 (b). The hammer at the instant of penetration:
51
PV Magazine “Insight”, online conference 14 October 2021, Presentation 6 “Rundown on thermal runaway” by Prof
Paul Christensen (Professor of Pure and Applied Electrochemistry, University of Newcastle)
https://www.youtube.com/watch?v=A9B5M8qHQQ0&t=3716s
28
Figure 3 (c). The hammer of 23 kg weight is thrown back into the air as the cells in the EV battery
module disrupt explosively:
Figure 3(d). A fraction of a second later, a black cloud of what is said by Prof Christensen to be
“cathode material” emerges.
It should noted that this black cloud potentially contains inhalable Nickel Oxides, a “Named
Dangerous/Hazardous substance” under Part 2 (Item 11)
52
with particularly low CQ or QQ in
Column 3 of the Schedule to the COMAH Regulations. Potential generation of just 1 tonne would
classify such installations as “higher-tier” COMAH if the chemical content is verified. See Tables 12
and 13.
52
https://www.legislation.gov.uk/uksi/2015/483/schedule/1
29
Figure 3 (e). Ignition of flammable gases follows in a fraction of a second later, with a large fireball
emitting considerable radiant heat:
Figure 3 (f). The “long, flare-like flames” last for several minutes as the fire progresses after the
initial explosion and ignition
30
3.2.2 Charged cells as Explosive Articles:
The video evidence certainly suggests, prima facie, that charged Li-ion cells should be regarded
as Explosive Articles in Division 1.2 (for the cylindrical cells with a projectile hazard) or Division 1.3
(for the stack of pouch cells) which certainly seems to correspond qualitatively, in the test
abstracted in Figure 2, to the description in Annex I Reg. 2.1.2.2(c) of the CLP Regulation
53
:
Division 1.3: Substances, mixtures and articles which have a fire hazard and either a minor blast hazard
or a minor projection hazard or both, but not a mass explosion hazard:
(i) combustion of which gives rise to considerable radiant heat; or
(ii) which burn one after another, producing minor blast or projection effects or both;
The subsequent behaviour in the Christensen field tests likewise corresponds qualitatively to a
Division 1.3 Explosive Article. These tests use mechanical penetration to initiate failure (as is done
elsewhere in the video
54
of the cylindrical consumer cells) but the other evidence makes clear that
either overheating, or overcharging
55
, are capable of initiating cell failure.
There are test procedures mandated in the CLP Regulation for assignment of substances,
mixtures or articles to the various Divisions of the Explosives class. These are referenced to the
United Nations Recommendations on the Transport of Dangerous Goods (UN RTDG) Manual of
Tests and Criteria
56
(MTC) published by the UN Economic Commission for Europe (UNECE) which
sets out clear standards of testing for assignment of articles or substances to the various Divisions
of Explosives. It is explicitly mandated by the CLP Regulation, as seen in many places, but explicitly
in Annex I Reg. 2.1.4.1
57
:
2.1.4.1.The classification of substances, mixtures and articles in the explosives hazard class and further
allocation to a division is a very complex, three step procedure. Reference to Part I of the UN RTDG
Manual of Tests and Criteria is necessary.
Section 16 of the MTC defines “Test Series 6” which is the critical set of tests for assignment of
explosives, as in the decision flowchart required by the CLP Regulation Annex I, Figure 2.1.3
58
,
distinguishing the various Divisions. It is not known whether tests according to this specification
have ever been explicitly carried out for BESS cells (as opposed to consumer product batteries) but
they are the tests mandated by the CLP Regulation.
For example Test 6 (c) is an “External fire (bonfire) test” with the stated objective “to determine
whether there is a mass explosion or a hazard from dangerous projections, radiant heat and/or violent
burning when involved in a fire”. The test package is to be not less than 0.15 m3 (150 litres) which is
significantly larger than the stack of cells shown in the test of Figure 2. The test package is placed
on a support grid above a wood, kerosene or gas fire which achieves a temperature of 800 °C, and
there are witness screens (2mm thick aluminium sheets) in four quadrants at 4 m from the edge of
the test packages.
53
https://www.legislation.gov.uk/eur/2008/1272/annex/I/division/2/division/2.1/division/2.1.2/division/2.1.2.2
54
Li-ion batteries video: 2.6 Ah consumer cells. https://www.youtube.com/watch?v=CUgbmCSmSNY
55
Li-ion batteries video: stack of pouch cells https://www.youtube.com/watch?v=EDhE0pk3FeQ
56
United Nations (2019). Manual of Tests and Criteria, 7th revised edition, ST/SG/AC.10/11/Rev.7 UN Publication Sales
No. E.20.VIII.1 ISBN 978-92-1-130394-0
https://unece.org/fileadmin/DAM/trans/danger/publi/manual/Rev7/Manual_Rev7_E.pdf
57
https://www.legislation.gov.uk/eur/2008/1272/annex/I/division/2/division/2.1/division/2.1.4/division/2.1.4.1
58
See e.g. Guidance on application of CLP criteria, version 5.0 July 2017, page 99
31
Allocation criteria on the basis of this bonfire test are given in section 16.6.1.4 of the MTC
59
.
Because the available evidence from Figure 2 is in a much smaller test packaged than specified in
the MTC, it is not possible to decide definitively how a specified test package (150 litres) would
behave in such a fire test. However the behaviour in Figure 2 already corresponds to the 1 m
fireball criterion of paragraph 16.6.1.4.5 (a) which would assign a 150 litre test package to Division
1.4. A standard size test package of 150 litres would be significantly larger and very plausibly would
exhibit a larger fireball passing criterion 16.6.1.4.4 (a) (fireball greater than 4 m distance) which
would assign the article to Division 1.3. The burning time or irradiance criteria of paragraph
16.6.1.4.4 (c) would also assign the article to Division 1.3 and from the behaviour in Figure 2 the
burning time is much shorter than the criterion’s 35 seconds, albeit for a smaller package than the
100 kg reference.
Without more detail on the cell types, sizes, energy stored for the cell stack shown in Figure 2, it
is not possible to say definitely how a test standard cell-stack (150 litres) would behave in the Series
6 Type (c) “bonfire” test of the MTC, and certainly not possible to decide quantitatively on the
criteria listed in section 16.6.1.4 which would discriminate between Division 1.3 and 1.4 Explosives.
Such determinations would require actual tests according to the specifications laid out in the
UN MTC. Because the behaviour is known to depend on the cell’s SoC, such tests should be
performed at several different SoCs, including 100% charged.
Although cells may be shipped and installed at lower SoCs than 100% (30% is the recognised
standard SoC for shipping consumer product Li-ion batteries) the cells in a BESS are designed to be
taken to high SoCs several times a day: that is the normal design operation of a BESS functioning for
grid-scale energy storage.
Hence such cells are likely to pass through states qualifying as Explosives in the Series 6(c) tests
of the MTC, several times daily.
In the absence of actual tests, the available evidence suggests that Li-ion cells should be
regarded as Explosive articles in Hazard class P1a (for Division 1.3 Explosive articles). Qualitatively,
the video evidence corresponds to the description in Annex I Reg. 2.1.2.2(c) for a Division 1.3
Explosive, presenting “a fire hazard and either a minor blast hazard …” and “considerable radiant
heat”. It does not correspond to the description of a Division 1.4 Explosive with “effects largely
confined to the package” .
A typical BESS composition (from an actual Planning Application document) is given later in
Table 7, with an inventory of active substances in the cells of 6.46 tonne per MWh of energy
storage. This will vary according to cell type and manufacturer, but may be taken as representative.
The 10 tonne CQ for the P1a Explosive category is exceeded for an energy storage of 1.55 MWh.
The 50 tonne CQ for the P1b Explosive category (which includes Division 1.4 Explosives) is
exceeded for an energy storage capacity of 7.75 MWh.
59
Page 171 of the MTC, 7th edition.
32
3.3 Routes to generation of Part 1,2 Hazardous Substances in loss of control accidents in BESS:
(II) Generation of H1 and H2 Acute Toxic gases in oxidising conditions
Toxic gases in Table 1 include not only the incidental hydrogen chloride HCl and hydrogen cyanide
HCN “typical of plastics fires” but also the notoriously corrosive hydrogen fluoride HF and the
unstable phosphoryl fluoride POF3. HF in particular can be evolved in large quantities, originating
from the large inventory of fluorine-containing compounds used in Li-ion battery cells. These
include the lithium hexafluorophosphate LiPF6 which is a critical ingredient of the electrolyte, and
PVdF (polyvinylidene fluoride or difluoride), a polymer typically used as a separator membrane.
Generation of HF and other toxic fluorine-containing compounds is therefore an unavoidable
aspect of Li-ion battery electrochemistry. These fluorine-containing compounds and polymers are
essential aspects.
HCN is a notoriously toxic substance though as a pure substance (a very low-boiling liquid) it
has a Harmonised Classification only as H2 Acute Toxic. In a mixture however the classification is H1
Acute Toxic, which aligns with the majority of the Notified Classifications in the C&L Inventory (H1
for all forms). The H1 classification is found for mixtures in the authoritative GB Mandatory
Classification and Labelling List.
HF has a dual classification as a H1 Acute Toxic gas by the dermal (skin attack) route, and also as
a H2 Acute Toxic by the oral and inhalation routes. By Note 7, the lower Controlled Quantity
applies, meaning a CQ of 5 tonnes in respect of HF.
POF3 has no listed toxicity assessment, probably because it is too unstable to perform the
testing. It is known to hydrolyse readily in water
60
generating three molecules of HF for each one of
POF3, so it is justifiable by Note 6 to assign a classification of H1 Acute Toxic (dermal), as it is a
known precursor of HF.
Finally, carbon monoxide CO has been considered in the previous section as a “Physical Hazard”
as it is a Flammable Gas classified as a P2 Physical Hazard. However it also a toxic gas, classified as
H2 Acute Toxic, with a CQ of 50 tonnes. Moreover, the available data on HF and POF3 were
obtained in oxidising conditions in a full “fire test”, and under oxidising conditions further evolution
of CO is expected from partial combustion not only of the flammable “P2” hydrocarbons considered
above, but also of the powdered graphite (carbon) used in large quantities as an anode material.
Evaluation of quantities and application of the Aggregation Rule for Health Hazards therefore
requires estimates of the likely quantities of HCN, HF, POF3 and CO generated in BESS fires.
Quantities of HCN: There is very little literature data, although concentrations of HCN were a
contributory hazard to fire fighters in the 2019 McMicken incident. HCN probably arises from
incidental nitrogenous polymers used in packaging or cell construction. For want of any other
estimate, we may take the report
61
of measurements at the time, where “above 50 ppm” (HCN)
and “above 500 ppm” (CO) were recorded. These are compounds with very similar molecular
60
Wikipedia entry for phosphoryl fluoride. https://en.wikipedia.org/wiki/Phosphoryl_fluoride
61
McKinnon, M B, DeCrane S, Kerber S (2020). “Four fire-fighters injured in Lithium-ion Battery Energy Storage System
explosion – Arizona”. Underwriters Laboratories report July 28, 2020. UL Firefighter Safety Research Institute,
Columbia, MD 20145
33
weights, so a reasonable indicative estimate would be to assume that HCN may be generated at
1/10 of the quantity of CO. The cell type at McMicken was NMC, so the estimates from the previous
section of 102 mg/Wh of CO are probably representative – at the “anoxic” stage of the incident
62
. A
rough estimate of 10 mg/Wh of HCN is therefore justifiable on these figures.
This can only be taken as an order of magnitude. Further assessment requires measurements of
HCN evolved in actual fire tests of representative cells complete with packaging, wiring etc.
Quantities of HF and POF3: Literature sources document the evolution of these gases in fire tests of
a variety of actual Li-ion cells. The leading source documenting HF and POF3 releases in open fire
tests is by Larsson
63
et al. who report HF releases from a variety of cell types but predominantly
“pouch” format cells of the LFP cathode type. The test set included LCO (lithium cobalt oxide)
cathodes and NCA-LATP (lithium nickel cobalt alumina – lithium aluminium titanium phosphate) but
no examples of the NMC (nickel manganese cobalt) cathodes included by Golubkov. The latter
evidence suggests that LCO and NMC cathode types are likely qualitatively similar, comparing
mixed LCO-NMC and pure NMC.
Independent HF release estimates from other fire tests are given by Sturk
64
et al. Many
experts
65
note that although LFP cathode cells are widely promoted as a “safer” technology
(because of the higher thermal runaway threshold (Table 3), the lower temperatures reached in
thermal runaway, and the generally slower “burn” in failure), the quantity of HF released in failures
is significantly higher than for the mixed oxide (NMC, LCO etc) cathode types. LFP cathode cells are
thus significantly more hazardous in terms of toxic emissions than NMC or other mixed oxide types.
Emissions vary according to State of Charge (SoC) and according to the packaging “form factor”
(i.e. whether pouch, cylindrical or prismatic), so once again this points to the importance of actual
fire tests on representative samples of the actual cells to be deployed in any installation. Quantities
abstracted from Larsson et al. and from Sturk et al. are summarised in Table 6.
Table 6: Fluoride gases generated from LFP cells compared to LCO or NMC cells from two sources.
62
There was no “fire” as such at McMicken, only a vapour cloud explosion. The “anoxic” conditions of the previous
section are therefore reasonable for this purpose.
63
Larsson F, Andersson P, Blomqvist P, Mellander BE (2017). Toxic fluoride gas emissions from lithium-ion battery fires.
Scientific Reports 7, 10018 doi: 10.1038/s41598-017-09784-z
64
Sturk D, Hoffmann L & Tidblad AA (2015) Fire Tests on E-vehicle Battery Cells and Packs, Traffic Injury
Prevention, 16:sup1, S159-S164, DOI: 10.1080/15389588.2015.1015117
65
Mrozik W, Rajaeifar M A, Heidrich O, Christensen P (2021) Environmental impacts, pollution sources and pathways of
spent Li-ion batteries. Energy Environ. Sci., 2021, 14, 6099-6121. DOI: 10.1039/D1EE00691F
66
Data from Table A4, online Supplementary Materials.
Gas
Larsson et al.
Sturk66 et al.
LCO prismatic
(25 Wh nominal)
LFP pouch cells
(64 Wh nominal)
NMC
(14 Ah taken as 50 Wh)
LFP
(7 Ah taken as 22 Wh)
(values in mg / Wh energy capacity)
HF
20 – 25
170 – 200
(max. at 0% SoC)
6.2 – 19
(max. at 100% SoC)
50 – 132
(max. at 100% SoC)
POF3
15 – 22
not detected
not measured
34
The Sturk quantities for NMC cells are somewhat lower than the results for the LCO cells
reported by Larsson, though Sturk does not report POF3 (a known HF precursor). The results for LFP
cells are also lower in Sturk than the highest values reported in Larsson. There is no consistency
regarding State of Charge. Generally speaking LFP cells behave differently from the other cathode
chemistries, for which behaviour is qualitatively similar.
The Larsson data are more thorough than in Sturk; the Sturk data require estimates of terminal
voltages (unstated) to derive energy capacities, and prudent conservatism in safety analysis
demands that highest reported values be taken, so we will take the Larsson values for LFP pouch
cells, with the LCO cells as a comparator, the type for which POF3 emissions are reported. The
comparison with Sturk shows that LCO cells are likely to be representative of NMC or other mixed
transition metal cathodes.
Quantities of CO: For estimation of CO evolution it is no longer sufficient to consider anoxic
conditions as in the Golubkov paper in the preceeding section. In an oxidising fire, all organic
components of the cells and the significant inventory of graphite (carbon) used in the anodes will
burn, although the degree to which combustion is complete will depend on circumstances. Also the
composition by mass of the cells proposed is required.
An outline composition for the chemicals content of a whole site of BESS cells of 26.3 MWh
capacity is given in a memorandum by D. Haigh of Golder Associates
67
(UK) Ltd written for a BESS
applicant (SYNERGY) in relation to a Planning Consent granted for a BESS at Kells, Co. Antrim,
Northern Ireland
68
. This gives a composition for the entire site abstracted in Table 7.
structural
component
Mass (tonnes)
per 26.3 MWh
tonne /
MWh
Cobalt oxide
CO
cathode active
materials
19.1
0.726
Manganese
dioxide
MnO2
19.1
0.726
Nickel oxide
NiO
19.1
0.726
carbon
C (graphite)
anode active
material
38.1
1.449
electrolyte
LiPF6 in
organic
carbonates
active electrolyte
28.6
1.087
PVdF
Polyvinylidene
fluoride
separator
9.5
0.361
Aluminium foil
Al
cathode support
19.1
0.726
Structural
aluminium
Al
housings
21
design-
dependent
Copper foil
Cu
anode support
17.2
0.654
Mass functional substances per MWh :
6.456
Energy density ( Wh/kg ) :
155
Table 7: Tonnages of active substances at an actual 26.3 MWh BESS in Northern Ireland
67
Attenborough House, Browns Lane Business Park, Stanton-on-the-Wolds, Notts NG12 5BL
68
Obtained by a Freedom of Information application to Antrim and Newtownabbey Borough Council, December 2021
35
Although the chemical technology is evolving continuously, the energy density for the cells
implied in Table 7, around 150 Wh/kg, is consistent with literature reports elsewhere. For
estimation of the potential for generating Carbon Monoxide, the Carbon-containing components
need to be identified. The organic solvents used in the electrolytes are a minority but significant
part of the total Carbon content, which is dominated by the graphite used in the anodes. The
Carbon content of the whole electrolyte can be estimated on the basis of typical compositions of
the solvents and concentrations of the LiPF6 salt found in the technical literature. The resulting
Carbon composition of the cells in mg / Wh is given Table 8.
substances
Mass Fractions
of Carbon in
component
(electrolyte, or
anode material)
Mass (mg / Wh)
of Carbon,
for BESS in Table 7
Organic solvents
in electrolyte
(dimethyl-, ethyl-
methyl-, ethylene-
and propylene
carbonates)
Mass fraction of
solvent in electrolyte,
1.2 mol L-1 solution69
Mass fraction70 of
Carbon in solvent
85.8 %
41.1 – 41.5 %
35.57 %
386.6
PVdF membranes
(CnHnFn)
37.5 %
135.4
Graphite anodes
-
-
100 %
1449
1971
Table 8: Estimated carbon content of Li-ion cells used in the BESS in Table 7.
Finally, the approach taken in the Golder memo was to estimate the CO2/CO ratio in combustion
in fire from actual fire tests published by the insurer F M Global
71
. The source is important as one of
very few sources to provide actual mass measurements in full scale fire tests of burning batteries.
Key parameters used in Golder are given in Table 9.
CO2 / CO mass
ratios
Equivalent CO2 /
CO mole ratios
Mole fraction CO
in gas mixture
C-content of
BESS anodes
“Worst case” CO
generation
43 – 71
27.4 – 45.2
3.52 – 2.16 %
38.1 tonne
3.1 tonne
Table 9: CO2/CO ratios in open fire tests reported by F M Global and application to Table 7 BESS
This will be representative of conditions in an uncontrolled fire with free supply of oxygen or air.
However this will not necessarily be the case if air supply is restricted by enclosure design,
conditions at the site, or especially by the deployment of so-called “clean agent” or AFC (aerosol-
forming composite) fire suppression systems which attempt to smother a fire and restrict air
69
Typical for Li-ion battery cells. See: Logan E R, Tonita E M, Gering K L, Li J, Ma X, Beaulieu L Y, Dahn J R (2018). A Study
of the Physical Properties of Li-Ion Battery Electrolytes Containing Esters. J Electrochem Soc 165(2) A21-A30 [DOI:
10.1149/2.0271802jes]
70
From various solvent compositions in the Golubkov study. Different compositions do not differ greatly in Carbon
mass fraction.
71
F M Global (2013) Flammability characterization of Li-ion batteries in bulk storage. Available at
https://www.fmglobal.com/research-and-resources/research-and-testing/research-technical-reports
36
supply. The assumptions of the Golder memo (taking the worst case CO2 / CO ratio in open fire
tests) cannot therefore be said to be conservative for the formation of Carbon Monoxide (CO).
The Golder assumptions and the lowest CO2/ CO mass ratio from the F M Global source provide
only a lower estimate of the potential for Carbon Monoxide generation in fire, for the purposes of
the HSC Regulations. The argument taken in Golder was the opposite, that the “worst case” CO2 /
CO was conservative because complete consumption of all Carbon containing components was
unlikely, and the “worst case” CO2 / CO ratio was derived from actual fire tests.
However this is to misunderstand the HSC Regulations, where, as argued above, it is the mere
presence on the site of controlled substances which are regulated; the likelihood of accidents and
of their extent is not relevant. For the purpose of the HSC Regulations, the substances “S” which are
regarded as hazardous in “loss of control” are any substances which may generate listed hazardous
substances (Parts 1 or 2) under loss of control. Carbon monoxide is certainly a Parts 1 and 2
hazardous substance which is known to be generated in loss of control situations; the regulated
“substances S” that may generate CO are the polymers, the organic solvents, and the graphite in
the anodes. As remarked above, it is readily conceivable that actual CO proportion might be worse
than the so-called “worst case” from F M Global, if air supply is restricted, for any reason. The
quantity derived from the figures in Table 9 is therefore not conservative, as claimed by Golder.
Both the Golder assumption and the complete conversion assumption are given in Table 10, for
the carbon containing components.
Source
of
Carbon
Mass of Carbon
in components
(Table 8)
CO generated:
as 3.52 mole %
of combustion
products
as 100 mole % of
combustion products
(mg / Wh)
tonne (26.3 MWh)
solvents
386.6
31.8
902
23.7
PVdF
135.4
11.1
316
anodes
1449
119
3381
totals
1971
161.9
4599
Table 10: Mass of CO generated from carbon-containing components of a 26.3 MWh BESS,
under two assumptions for mole fraction of CO in combustion products.
The result for 100% conversion of the organic solvents in the 26.3 MWh BESS in Table 7 is
shown explicitly, because this assumption appears to have been made in Golder, for the organic
solvents only
72
, whilst the 3.52% mole fraction used for the combustion of graphite.
72
The figure in Table 2 of Golder is 22.5 tonne CO generated from the organic carbonates only. The difference from the
23.7 tonne in Table 10 is readily explained by different assumptions (or knowledge) for the electrolyte composition
conjectured in Table 8.
37
For the purposes of this paper, shall take the 3.52% mole fraction of CO in the carbon oxides
combustion gases as a rational minimum CO generated in fire situations. This is not necessarily
conservative as argued above.
Because 100% conversion of organic solvents to CO has been used in one evaluation of CO
generation from a BESS fire (by consultants acting for an Applicant), we shall take this as a
conjectural upper figure for estimation purposes. As before, further evaluation can only come from
actual fire tests under various representative conditions of air supply or partial smothering.
Toxic gas generation from BESS in fire situations is summarised in Table 11. LFP cells are
contrasted with “Mixed Oxide” cells taking data for the LCO cells in Table 6, and for a range of
estimated CO quantities derived from the preceeding considerations in Table 10. For the lower
limit, we take 3.52 mole % of CO in combustion gases for a typical inventory (Table 7) of carbon-
containing materials. For the upper limit, we use the figure assuming 100% conversion of the
organic solvents to CO (Table 10), as used in one prior BESS fire assessment submitted to a Planning
Authority (Golder). As discussed above (Table 6) the fluoride gas data from Larsson for LCO cells
was the first to report POF3. Generally LCO cells appear to be similar to NMC and other mixed oxide
cathode types in their breakdown behaviour, in contrast to the qualitatively different LFP
behaviour, especially for fluoride emissions.
As in Table 5, toxic gas generation is scaled to a Reference Case 50 MWh of energy storage, and
their contributions to the summation in the Aggregation Rule (for Health Hazards) are shown in the
right-most columns, for the two cell types. The energy storage at which the Aggregation Rule sums
to unity is also given, in MWh, using both ends of the range given for CO.
We conclude from the data given in the cited sources, that BESS based on Mixed Oxide cathodes
would trigger a HSC assessment at a storage capacity between 34.5 – 68.5 MWh, based on the
Aggregation Rule. For a BESS based on LFP cells, the assessment threshold is significantly lower,
between 16.7 – 22.1 MWh. The range is also smaller, as the estimates for LFP cells are dominated
by the fluoride gases, whereas the estimates for the LCO cathodes are more sensitive to the
assumptions made regarding CO.
Moreover, the LFP cells exceed the CQ for H1 acute Toxics on the basis of HF alone, at 25 MWh,
without requiring any assumptions regarding completeness of combustion in generation of CO.
We remark under Table 5, that further accuracy in the “Physical Hazards” assessment could only
be obtained from actual closed container (anoxic conditions) tests on representative samples of the
actual cells to be installed. Similarly, we could only obtain further accuracy in the Toxic Gases
assessment, by actual fire tests on representative samples of the cells to be installed, under a range
of scenarios regarding air supply (which could affect the mole fraction of carbon monoxide in
combustion products).
38
substance
HSC Hazard
Category
Controlled
Quantity
Masses ( mg / Wh )
(equivalent kg/ MWh)
Masses
(tonne / 50 MWh)
Contribution to
Aggregation Rule
(tonnes)
LCO
LFP
LCO
LFP
LCO
LFP
HCN
H1 Acute
Toxic
5
10
10
0.5
0.5
0.1
0.1
HF
H1 Acute
Toxic
(dermal)
5
25
200
1.25
10
0.25
2.0
POF3
H1 Acute
Toxic
(provisional,
Note 6)73
5
22
-
1.1
-
0.22
-
CO
H2 Acute
Toxic,
Category 3,
inhalation
50 74
162 – 902 75
8.1 – 45.1
0.162 – 0.902
Aggregation Rule sums for 50 MWh Reference Case:
0.73 – 1.5
2.26 – 3.0
Storage (MWh) at which Aggregation Rule for Health Hazards > 1 :
(upper case for CO generation)
34.5 MWh
16.7 MWh
Storage (MWh) at which Aggregation Rule for Health Hazards > 1 :
(lower case (3.52 mole % CO) for CO generation)
68.5 MWh
22.1 MWh
Table 11: Toxic gases generated from BESS in oxidising conditions for two contrasting cell types,
and a range of quantities of CO generated from carbon-containing components.
73
Larsson et al. comment: “Judging from its chlorine analogy POCl3/HCl, POF3 may even be more toxic than HF”.
74
As H2 Health Hazard rather than as P2 Physical Hazard, where the CQ is 10 tonne.
75
From Table 10, boldface figures. Note maximum CO generated in anoxic conditions (Table 5) is 218.8 mg / Wh,
reported for LCO-NMC cells by Golubkov.
39
3.4 Routes to generation of Part 1,2 Hazardous Substances in loss of control accidents in BESS:
(III) inhalable Nickel compounds.
A final consideration for Li-ion BESS using some mixed oxide cathodes containing nickel oxides is
the possibility of generating dust or smoke containing “Nickel compounds in inhalable powder
form”, which are Named Hazardous Substances in Part 2 of Schedule 1 to the HSC Regulations
(entry number 11)
76
. The inhalable powders referred to specify the oxides and sulphides of nickel
listed in Table 12.
Clearly these cannot be a concern if Nickel oxides are absent from the cell types deployed, but
NMC cells and LCO-NMC mixed oxide cells do contain Nickel, and the BESS in Table 7 is stated to
contain an inventory of 19.1 tonnes nickel oxide NiO for 26.3 MWh storage capacity
77
.
The requirement for this to become a Named Hazardous Substance in Part 2 is that they be
present in “inhalable powder form”, so the question arises of release of nickel oxides in fires, as
inhalable dust or smoke. The form in which they occur in the Li-ion cells is already nanoporous; all
that would be needed is release from the aluminium foil support with mechanical disruption in
thermal runaway accidents.
The question is an important one for health protection, and the operation of the HSC
Regulations, because the Controlled Quantity is particularly low, only 1 tonne.
The total inventory of NiO in a modest-sized BESS (26.3 MWh) in actual practice, as reported to
a Local Planning Authority, is already 19.1 tonnes (see Table 7), well in excess of the Controlled
Quantity.
If it is “reasonable to foresee” the “fixed” form of NiO within the Li-ion cells becoming released
in “inhalable powder form” e.g. in fires or thermal runaway events, then even quite small BESS
(around 1.4 MWh capacity) may require HSC, by exceeding the CQ of a Part 2 Named Hazardous
Substance. Moreover, the parallel entry in the COMAH Regulations appears in Column 3 of the
Schedule to the COMAH Regulations
78
, so such a site would be classified as an “upper-tier” COMAH
site as soon as the threshold is exceeded, without any “lower-tier” categorisation.
Compound:
formula
Hazard Codes
Controlled Quantity
(Named substance, Part 2)
nickel monoxide
NiO
H350 carcinogen by inhalation
1 tonne
nickel dioxide
ONiO
H350i
dinickel trioxide
Ni2O3
H350i
nickel sulphide
NiS
H350i, H400, H410
trinickel disulphide
Ni3S2
H350i, H400, H410
Table 12: specified Nickel compounds that are Part 2 Named Hazardous Substances if present in
“inhalable powder form”, being known or suspected carcinogens by inhalation
76
https://www.legislation.gov.uk/uksi/2015/627/schedule/1/made
77
This was the reference case for the cited memo. The actual BESS in development is about double this size.
78
https://www.legislation.gov.uk/nisr/2015/325/schedule/1
40
3.4.1 Evidence for the generation of inhalable Nickel Oxides in loss of control accidents
Whether the generation of nickel oxides from Li-ion battery cells is possible depends on the
cathode chemistry type. Clearly, if nickel is not a component of the Li-ion cell, then significant
generation of nickel oxides is impossible. However the NMC and NMC-LCO cell types do contain
Nickel, many BESS already installed use the NMC type, and in at least one Planning application
79
in
process, the developer has stated that the cells may be either LFP or NMC.
Evidence for the generation of inhalable nickel oxides comes from the video records of
controlled failures made by Christensen already referred to in Section 3.2.1 above.
Figure 3 (d) (page 27) shows the stage of an explosion of a 1.7 kWh module destroyed in a field
test where a cloud of initially black smoke emerges from the disrupting module. In this test the
ignition of flammable gases followed immediately upon the explosive disruption in a “deflagration
event”.
Other experiments by Christensen are shown in the same lecture source
80
in which the ignition
is not in fact immediate. Christensen summaries the possible course of events in the “flowchart” of
Figure 4 below. Whether the ignition of flammable gases and vapours is immediate or delayed
depends on circumstances, notably the relative concentrations of flammable gases and air (or
oxygen). Delayed ignition (likely to encouraged, not inhibited, by aerosol-based fire suppression
systems) is in many ways more dangerous, as no flame need be involved until an explosive mixture
is generated, leading on contact with a hot surface to a Vapour Cloud Explosion.
Figure 4. Summary of possible courses of events in a BESS “thermal runaway” event. From lecture
by Prof Paul Christensen.
79
Letter from Sunnica to Mrs Lucy Frazer QC MP
80
PV Magazine “Insight”, online conference 14 October 2021, Presentation 6 “Rundown on thermal runaway” by Prof
Paul Christensen (Professor of Pure and Applied Electrochemistry, University of Newcastle) https://www.pv-
magazine.com/pv-magazine-events/insight-australia-2021/
41
An example of a thermal runaway event without immediate ignition, also from Christensen’s work,
is shown in Figure 5, where the concentration of flammables in air was such that the mixture never
ignited during the test
81
. This parallels the situation known to have occurred in the heavily-
analysed
82
explosion at McMicken, Arizona in 2019, where no flame was involved until a
“deflagration event” which left first-responders with life-limiting injuries. Various stages in the
generation of a vapour cloud are shown in panels (b) (c) and (d). Figure 5(a) repeats Figure 3(d) for
comparison of the “black cloud” stage.
Figure 5(a). Same as Figure 3(d), showing the early emergence of a “black cloud” before a “white
cloud” likely to be dominated by droplets of organic solvents. In this test the ignition followed
shortly thereafter.
Figure 5 (b). Initial failure of a 1.67 kWh module in a low State of Charge (SoC) (40%). The black
cloud (said by Christensen to be “cathode material”) emerges in the first fraction of a second.
81
PV Magazine “Insight”, online conference 14 October 2021, Presentation 6 “Rundown on thermal runaway” by Prof
Paul Christensen (Professor of Pure and Applied Electrochemistry, University of Newcastle)
https://www.youtube.com/watch?v=A9B5M8qHQQ0&t=3716s. Later in same lecture.
82
Hill, D. M. (2020). McMicken Battery Energy Storage System Event: Technical analysis and Recommendations. Report
from DNV-GL to Arizona Public Service. Doc. No. 10209302-HOU-R-01. 18 July 2020,
42
Figure 5 (c): The white vapour cloud follows a fraction of a second later. The light-scattering
material creating the opaque white material is likely to be the organic solvents from the cell.
electrolyte.
Figure 5 (d). The vapour cloud eventually fills the container, without ignition, during the test.
An example of vapour cloud formation before an actual BESS explosion
83
(the September 2020
Liverpool explosion and fire) is shown in Figure 6 from actual CCTV records shortly before the
explosion. A black cloud analogous to Figures 3(d) and 5(a) is not captured, but as seen from the
recordings this is typically an initiating event rapidly superseded by the emergence of a white cloud.
83
Merseyside Fire and Rescue Service, Fire Investigation Report 132-20, Incident Number 018965, Carnegie Rd,
Liverpool, September 2020. Report dated February 2022.
43
Figure 6 (a). From internal CCTV, seconds before the Liverpool explosion. Figure 15 of report.
Figure 6(b). The vapour cloud fills the local region of the container before ignition. Figure 16 of
report.
44
Christensen states at several points (including the notes in Figure 4) that the “black cloud” seen
very early in Figures 5(a) and 5(b) is “cathode material” and likely to consist of toxic heavy metal
nanoparticles, including manganese, cobalt and nickel (which are all used in the cathode, in NMC
cells). The occurrence as nanoparticles is not surprising since the cathode material is already
nanoporous in its deposition.
The nanoparticles emitted are likely to include oxides, since the cathode material in so-called
“NMC” cells is a mixed nickel-manganese-cobalt oxide. Metallic oxide nanoparticles (MO-NP) form
readily and nanoporous MO are already present in the cathodes. Metallic nanoparticles (MNP) can
be stable and might also be present, though Christensen confirms
84
that the oral description of
“toxic heavy metal nanoparticles” should have been of “metal oxide nanoparticles”.
As smoke, this is patently an “inhalable” form, likely to survive as smoke or dust in any
subsequent fire or explosion. If Christensen is correct, that the “black cloud” emerging in these
failure records contains nanoparticles of the “cathode material” (the component containing Nickel
in NMC cells), then it is likely that these images demonstrate, for NMC and related cathode types,
the presence of “Nickel Oxides in inhalable powder form” which is a Named Hazardous Substance
in Part 2 of the Schedule to the HSC and COMAH Regulations.
The chemical composition of the vapour clouds (both “black” and “white”) emerging in failure of
NMC and other Nickel-containing cell types is therefore important from a regulatory perspective,
and should therefore be definitively confirmed. If the “black smoke” is confirmed to contain Nickel
Oxides, then Nickel-containing cell types would probably require Hazardous Substances Consents at
energy storage capacities as low as 1.4 MWh.
3.4.2 Other literature evidence for inhalable Nickel Oxides from NMC Li-ion failure.
Multiple literature sources in fact confirm the likely presence of Nickel Oxide particles, or
particulates containing complex mixed oxides including Nickel, and at inhalable particle sizes.
Bergström et al. claim
85
to be the first (2015) to analyse “soot” or particles collected from smoke
generated in cell failure, detecting nickel compounds (by ICP-MS) from NMC cells.
Recent (2020) peer-reviewed literature
86
concludes that the ejected powders consist of carbon,
organics, carbonates, metal and metal oxides i.e. including Nickel oxides. Although the majority
component of the collected “soot” is carbon, these were closed cell tests gathering all particulates
generated, which could well have included graphite from the anodes. The same source (Chen,
2020) details the chemical decomposition of the NMC cathode into first Ni2O3 and then NiO (both
specified in Table 12) with evolution of free oxygen, an exothermic reaction explained elsewhere
87
in the context of fire behaviour dependent on State of Charge. The earlier (Ouyang, 2018) source
details how the NMC cathode decomposes first into the nickelous oxide Ni2O3 and the analogues
84
Christensen, P. June 2022. Private communication with EJF.
85
U. Bergström et al (2015), “Vented gases and aerosol of automotive Li-ion LFP and NMC batteries in humidified
nitrogen under thermal load” https://www.msb.se/siteassets/dokument/publikationer/english-publications/vented-
gases-and-aerosol-of-automotive-li-ion-lfp-and-nmc-batteries-in-humidified-nitrogen-under-thermal-load.pdf
86
S. Chen, Z. Wang and W. Yan (2020) “Identification and characteristic analysis of powder ejected from a lithium ion
battery during thermal runaway at elevated temperature”, J. Haz. Mat., 400,
123169 https://doi.org/10.1016/j.jhazmat.2020.123169
87
Ouyang, D., Chen, M & Wang J. (2018) Fire behaviours study on 18650 batteries using a cone calorimeter. J. Thermal.
Analysis and Calorimetry Doi: 10.1007/s10973-018-7891-6
45
Mn3O4 and Co3O4 from the manganese and cobalt content, with release of oxygen, followed by
further reduction to NiO, MnO and CoO with yet further release of oxygen, all dependent on
internal distribution of Lithium ions, and the higher to SoC, the more oxygen is generated internally,
explaining the increasing violence of “fires” with increasing SoC.
Other 2020 papers
88
analyse the collected powders finding particles with “huge amounts of
Nickel” and confirm that with the majority of particles are smaller than 10 µm2 (measured as cross-
sectional area in SEM images) they “can therefore be inhaled deeply into the lungs”. Detected
elements included Al, F and Ni and the particles are stated to be “carcinogenic and respirable for
humans”. The same source also details modes of failure, depending on State of Charge, and
estimates the mass loss from failed cells undergoing thermal runaway. The particle size analysis in
this paper (Essl, 2020) confirms unambiguously the generation of nickel-containing particulates in
“inhalable powder form”, meeting the description of Item 11 of the Part 2 “Named Hazardous
Substance”.
88
Essl, C., Golubkov, A. W. et al. (2020) Comprehensive hazard analysis of failing automotive Li-ion batteries in
overtemperature experiments. Batteries, 6,30 DOI: 10.3390/batteries6020030
46
More recent (2022) studies
89
confirm that the soot generated from EV fires consists of “metal
oxides of the cathode material, lithium and fluoride compounds”. The same source concludes that
NMC cells in thermal runaway produce “soot consisting mainly of heavy metal-oxides of Nickel,
Manganese and Cobalt (each 18-20% by mass)”. These results are from deposition of “fire soot” in
tests conducted in enclosed spaces mimicking vehicle parking facilities. These deposits were found
to consist largely of the metal oxides. The carbon soot reported in Chen (2020) was not found at
distance, presumably because these were actual fire tests in oxidising conditions, where fine carbon
dust would be expected simply to burn to CO or CO2. Moreover this “heavy metal oxide soot” is
transported over long distances (up to 100 metres) and deposited on surfaces. The same metal
oxides as in the dry powder were also found dissolved (or dispersed) in fire sprinkler water.
This paper has not carried out a comprehensive literature review, but the above sample of
sources suggests that, over the period 2015 to 2022, a rapidly accumulating body of knowledge of
chemicals released during Li-ion battery cell failure has emerged. In addition to the gaseous
substances detailed in Sections 3.1 (Flammables) and 3.3 (Toxics), substances generated in Li-ion
thermal runaway accidents definitely do include respirable dusts, transportable over distances of
100 m, definitely inhalable, and do include Nickel Oxides where the originating cell type contains
Nickel (typically NMC, a common chemistry in the EV industry). Essl (2020) clearly states that
“Particles should be considered as additional toxic hazard” (i.e. additional to Flammable and Toxic
gases), and moreover points out that the overall mass loss of cells tested (44%) comprises 74 g of
vent gases, but 300 g particulates (35% of original cell mass). Hence particulate emissions are in fact
dominant by weight.
Quantities of respirable Nickel Oxide dusts generated in battery failures remain unclear. The
emerging literature reports many different situations and measurement techniques, and is often
inconsistent. Estimates may be possible from mass loss measurements documented by Essl et al.
(2020), who found that mass loss in the form of collected particulates comprised 35% of the original
cell mass. The total electrode mass (cathode plus anode plus Al collector foil) in the cell comprised
68.5% of the original, suggesting that around half of the electrode material is emitted in the form of
particulates. Because some of the anode graphite may be expected to burn, cathode material
(43.5% of the original mass) is likely to predominate, though the particulates analysis in Essl (2020)
was not quantitative for composition.
An a priori estimate of 50% cathode material being lost in the form of inhalable dusts is
therefore not unrealistic, and may not be conservative i.e. conceivably larger fractions of the
cathode oxides may emerge as respirable dusts, smoke, or powders. Taking the data for an actual
commercial BESS in Table 7 (0.726 tonne/MWh NiO) , a “worst case” generation of respirable dust
would be 100% of the NiO quantities listed, with 50% a credible first estimate of particulates
generation based on Essl (2020). This would mean generation of inhalable NiO dust of about 0.363
tonne/MWh. The Controlled Quantity of “reasonably foreseeable” generation of the Part 2 Named
Hazardous Substance (CQ of 1 tonne) is breached at an energy storage capacity of 2.75 MWh.
This implies that even a single container BESS of the NMC cell chemistry would require HSC,
based on exceeding the Controlled Quantity of a Part 3 Hazardous Substance (the cells), generation
of over 1 tonne of a Part 2 Named Hazardous Substance being “reasonable to foresee in loss of
control of the processes”.
89
Held, M., Tuchschmid, M. et al. (2022) Thermal runaway and fire of electric vehicle lithium-ion battery and
contamination of infrastructure facility. Renewable and Sustainable Energy Reviews 165, 112474 DOI:
10.1016/j.rser.2022.112474
47
3.5 Routes to generation of Part 1,2 Hazardous Substances in loss of control accidents in BESS:
(IV) Generation of E1 Hazards to the Aquatic Environment in firefighting operations
After Physical and Health Hazards, the remaining Hazard Class to be considered is that of
substances hazardous to the aquatic environment in Category “Acute 1” or “Chronic 1” (E1
Environmental Hazard, CQ 100 tonnes) or in Category “Chronic 2” (E2 Environmental Hazard, CQ
200 tonnes).
Specific examples of substances that may be generated in fire or thermal runaway have been
discussed already in previous sections and listed in Table 2, which is not exhaustive. Transition
metal compounds in general, and compounds of copper, cobalt and nickel in particular, are all
known or potential toxins in the context of the aquatic environment.
The quantities estimation required by the HSC Regulations under Hazard classes E1 or E2
requires a minimum quantity of 100 tonnes; however this easily generated in fire-fighting
operations where water is used as an extinguishant (the only extinguishant recommended for
thermal runaway in Li-ion batteries).
A quantity 100 tonnes of contaminated water is readily conceivable in fire-fighting: this is
equivalent to 100 m3, or 100,000 litres, or 26,420 U.S. gallons. A rough marker is that a Tesla Model
S (battery capacity 100 kWh) that caught fire in Houston, Texas in April 2021 reportedly required
30,000 U.S. gallons to control, already in excess of the 100 m3 threshold for an E1 Environmental
Hazard, for a fire in a battery miniscule in size compared to the capacity of grid-scale BESS.
So it is almost inevitable during effective fire-fighting a quantity of potentially contaminated
water in excess of the CQ for a E1 Hazard will be generated “if control of the processes is lost”.
The other requirement is that the contaminated fire water falls into the Categories “Acute 1” or
“Chronic 1” (for toxicity to the aquatic environment) or in Category “Chronic 2” (for an E2 Hazard).
As argued above, this is again readily conceivable if copper oxides are generated from the anode
foils in fires; these oxides are conceivable from all current Li-ion battery types (even LFP) and carry
a M-factor of 100 (Harmonised Classification, either oxide, see Table 2) for the “Acute 1” hazard.
This would be the governing factor determining the concentration in contaminated fire water
that would classify such fire water as a E1 hazard; applying the M-factor and the “Summation
Method”
90
then under the rule in Table 4.1.1 of Annex I to the CLP Regulation
91
the contaminated
fire water is classified as “Acute Category 1” if the concentration of copper oxides dispersed in the
fire water exceeds just 0.25% by weight, requiring just 250 kg of copper (I or II) oxides in 100 m3 of
fire water, to exceed the Controlled Quantity of an E1 Environmental Hazard.
With a typical BESS containing 654 kg copper foil per MWh of energy storage (Table 7), this
could burn to create 736 kg copper (II) oxide, per MWh energy storage. This is already (at just 1
MWh), well in excess of the quantity required to create a E1 Environmental Hazard from fire water.
90
Annex I to the CLP Regulation, section 4.1.3.5.5. See Guidance document
https://echa.europa.eu/documents/10162/23036412/clp_en.pdf/58b5dc6d-ac2a-4910-9702-e9e1f5051cc5
91
Rule defined in Table 4.1.1 of Annex I to the CLP Regulation, also in the above Guidance document.
48
Similarly any generation of Cobalt compounds would, under the Summation Principle, further
reduce the concentrations of individual toxic compounds that would classify the resulting mixture
as “Acute” or “Chronic” Toxic to the aquatic environment, in Category 1.
These considerations show that detailed appraisal of the possibility of generating E1 or E2
Environmental Hazards in fire water is warranted; the Controlled Quantity is readily exceeded by
relatively modest fire-fighting operations.
The data that appear to be lacking for full application of the Regulations are data on those
transition metal compounds classified as toxic to the aquatic environment, their generation in fire
or thermal runaway accidents tackled with water, and their aquatic eco-toxicology (including M-
factors) where not already listed in the ECHA databases or the parallel UK databases.
These data are required first of all for the operation of the Summation Principle in the CLP
Regulation. This should not be confused with the “third application” of the “Aggregation Rule”
(Note 5) required under the HSC Regulations, which in this instance would “aggregate” quantities of
E1 and E2 Hazards below the Controlled Quantities (there are no Part 2 Named Hazardous
Substances in the Environmental Hazard category that are relevant to BESS).
Although the application of the Note 5 Aggregation Rule is in principle required, it is likely that
most fire-fighting operations would use in excess of the Controlled Quantity in any case, hence the
Aggregation Rule is unlikely to be the controlling factor from the perspective of the HSC
Regulations; the summation Principle of the CLP Regulation is in this case more important, for
classifying a complex mixture of compounds potentially present in the fire water.
It would appear however that the quantities of copper, cobalt, or nickel compounds likely to
classify contaminated fire water as “Acute Toxic Category 1” are quite small, and well within the
potential to be generated from typical compositions of Li-ion BESS in fires or thermal runaway. For
example, copper oxides at a concentration of just 0.25% (w/w) are an “Acute Toxic Category 1”
contaminated fire water. This is readily conceivable.
A precautionary approach requires the assumption that such toxic compounds will be generated
in fire, and any fire-fighting operation using in excess of 100 m3 fire water will then exceed the CQ
for a E1 Environmental Hazard.
Because these consequences are “reasonable to foresee if control of the processes is lost”, the
copper foil content of Li-ion BESS should be regarded as a Part 3 hazardous substance, and
potentially the cathode oxides in mixed oxide Li-ion battery chemistries.
One must again remark that actual fire tests, with water extinguishing and chemical analysis of
the resulting fire water, and establishing the eco-toxicology of compounds found, are required to
refine the above considerations.
49
4.0 Conclusions:
The significance of Table 5 (Physical Hazards) and Table 11 (Health Hazards) for the purposes of
evaluation requirements for HSC are compared in Table 13, which summarises the threshold in
energy storage capacity at which a HSC assessment is likely to be required, under various cell types
and assumptions regarding CO generation in fire. Included are the conjectured thresholds for
requirements based on inhalable nickel compounds, and on generation of contaminated fire water,
for which further information is required for the assignment of fully evidence-based thresholds.
The central conclusion of Table 13 is that a 50 MWh BESS is almost certain to require a HSC
assessment, regardless of electrode type or the assumptions made regarding CO. LFP cells are
widely promoted as “safer” than other chemistries because of their “slower” behaviour in thermal
runaway, but generate larger quantities of toxic fluorides. At 25 MWh, they are likely to require
HSC on the basis of HF generation alone, irrespective of assumptions regarding CO. NMC or other
mixed oxide cathodes may generate smaller quantities of toxic fluorides but including CO may still
trigger the Aggregation Rule on Health Hazards, and are almost certain to trigger the Aggregation
Rule on Physical Hazards, derived from anoxic conditions, similarly requiring no assumptions
regarding completeness of combustion.
The generation of inhalable nickel oxides, now widely reported in the technical literature,
appears to impose a particularly stringent threshold on BESS using nickel-based cathodes i.e.
complex oxides such as the NMC, NMC-LCO and NCA chemistries. The Controlled quantity of
inhalable Nickel Oxides, a Part 2 Named Hazardous Substance, is only 1 tonne, and this is also the
Qualifying Quantity for Upper-tier COMAH (there is no “lower tier” for this Part 2 Named
Substance). Based on a realistic (reasonable to foresee) estimate of 50% of the tonnage of NiO in an
exemplar commercial BESS, we estimate Controlled Quantities being exceeded for the Nickel-
containing cathode types at just 2.75 MWh, well within the capacity of a single container BESS. No
progression of a thermal runaway accident to adjacent containers is required for the CQ to be
exceeded.
For the smaller BESS, of non-Nickel cathode chemistries, there may be credible margins of
uncertainty. For BESS installations of 50 MWh or above, it is almost inconceivable that one or other
of the thresholds in Table 13 would not apply. Therefore, BESS at 50 MWh energy storage are
almost certain to require HSC, irrespective of cell type, and disregarding thresholds dependent on
an uncertain CO2/CO ratio in fire.
As remarked at multiple points, these thresholds are based on evidence available in the
published literature for representative cell types and conditions. Controlled and independently
verified tests on representative samples of actual cells proposed to be installed would be necessary
to establish legal thresholds more thoroughly than the a priori estimates set out here. These tests
should include (i) closed container tests to establish quantitatively generation of P2 Flammable
Gases (ii) open fire tests to establish H1 Acute Toxic fluorides, HCN, HCl and especially CO, under
different degrees of oxygen supply or deprivation, and (iii) fire tests with chemical analysis of
smoke, dust and contaminated fire water to establish Health and Environmental hazards from
nickel oxide dusts or contaminated fire water more precisely than can be done at present (iv)
consideration of the tests for “Explosive article” behaviour mandated in the UN MTC to determine
of BESS cells in a high SoC behave as Division 1.3 (Hazard class P1a) or Division 1.4 (Hazard Class
P1b) explosive articles.
50
Storage capacity
estimated at threshold
cell cathode type and
conditions
governing reason for HSC
1 MWh (text, E1 Hazards)
any, fire conditions with
water extinguishant > 100 m3
CuO-contaminated fire water exceeds
CQ of E1 Environmental Hazard
1.55 MWh (text, Explosives)
any (with adjustment for energy
density), at high SoC
Active materials exceed CQ for P1a
Explosive Articles
2.75 MWh (text, Sect. 3.4.2,
re Table 12 )
Nickel oxide cathodes (NMC,
or NMC-LCO, or NCA), in fire,
generating smoke or dust
NiO in “inhalable powder form”
exceeds CQ of Named HS in Part 2
7.75 MWh (text, Explosives)
any (with adjustment for energy
density), at high SoC
Active materials exceed CQ for P1b
Explosive Articles
16.7 MWh (Table 11)
LFP, high case for CO
Aggregation Rule > 1 for Health
Hazards (H1 or H2 Acute Toxic Gases)
22.1 MWh (Table 11)
LFP, low case for CO
25 MWh (text, re Table 11)
LFP, HF alone
HF exceeds CQ as H1 Acute Toxic
28.3 MWh (Table 5)
LCO-NMC, anoxic conditions
Aggregation Rule > 1 for Physical
Hazards (P2 Flammable Gases)
34.5 MWh (Table 11)
LCO-NMC, high case for CO
Aggregation Rule > 1 for Health
Hazards (H1 or H2 Acute Toxic Gases)
45.7 MWh (text, re Table 5)
LCO-NMC, anoxic conditions
CO exceeds CQ as P2 Flammable Gas
Table 13: Summary of energy storage capacity thresholds below 50 MWh likely to trigger a
requirement for HSC, for contrasting cathode chemistries and CO assumptions.
Italicized entry requires further data on behaviour of Li-ion cells in fire and of eco-
toxicology of contaminated fire-water, for a fully evidence-
based threshold.
