Incident Report

D i s t r i b u t i o n

Harpur H i l l , Wurton, SKI7 9JN

Audibility of the Public Fddress System a t Hillsboraqh FootbdLl Grrxlnd

by

I R Price, BSc MIOA MSEE

WLJNV/89/17

The Court of Inquiry (5 copies) ACC M Jones W e s t Midlands Police Mr D C T Eves Dr J -id Dr A Jones M r A B a r r e l l TD M r P G Jones TD Mr M S N a t t r a s s FAID Area 14 Mr J P G i l t r o w HFS (N) M r M R S t e p h e n s o n NE= M r M Fountain TD3 Dr J H Foley D r C E N i c h o l s o n Mr P F H e y e s M r I R Price M r D Waterhouse M r G A C G a m e s M r G N o r t o n M r M James m3 M r C J P e r t e e NEFCG M r J B H i b b s m m RFS Library (2) Registry Fi le

Issue authorised by: D r A Jones Date: 12 January 1990

-RLSD/DIAS:NurTO BE KMMWUCATED OUEIDE HSE w l x H c u T ~ A P ~ O F T H E m R I s x o m m

As part of the investigation undertaken by the Health and Safety Executive into the incident at Sheffield Wednesday FCfs Hillsborough football ground on the 15th April 1989, I was asked by Dr C E Nicholson to investigate the audibility of the Fublic Address sys tem (PA system) at Hillsbo~. This report sets out the results of that investigation. Information on the design of the PA system and the equiment used in it was supplied by Mr R Annible (Contracts Director) of DP S a d s y skns Litd at a meeting on the 8th June 1989 and in skeqyent comespndence and telephone conversations.

This report contains details of investigations perfonrsd bath on site and in the laboratory. The work was carried out by me and by other menkers of the Noise and Vibration Section of the Research and Laboratory Services Division (RLSD) working on my instruction and under my general supervision.

I was inform& by Mr Annible that the PA system at Hillsborou#~ was designed and installed by DP Sound Systems L W of Sheffield. ?he principle canpnents, lorrdspeakers and anplifiers, were made by TOA Electric CO Ltd, a Japanese ccanpany which is a specialist manufadxmr of public address equipment of same 55 years standing.

Mr Annible informed me that the PA system was installed primarily as a means of cammicating with the spectators for cmid control or in an emeqency. The system is controlled from the police control booth located between the West and South Stands. In the booth is a control unit (TOA VR 1012) on which the micrqhone input to the system is located together with a button which activates the system when depressed. The control unit also has a series of buttons which enable sectors of the graund to be selected so that messages can be relayed to specific areas (see Figure 1). The PA system has been divided into eight sectors; inside and outside of each of the four stank (Figure 2). The outside sectors also include the passages leading to the upp=r seated accammodation in each of the Any cathination of the eight sectors can be selected or, using the button labelled \AIL C . ' every sector can be selected. ?he sector outside the West stand also includes lcezdspeakers direckd at the approach to the turnstiles at Leppings Lane. Thus it is possible for the police to talk to those inside the gmund independently of those approaching it or vice versa. It is also, for example, possible to talk to those inside and/or approadhg the West staml/temacing irdependently of those anywhere else in or around the ground-

A second input to the PA system has been provided for use by a Disc Jockey (DJ) to play music and jingles to the spectators and to make -ts. This inpt uses part of the same PA system, that directed at all spectators inside the grcrcmd, ie the four sectors inside the stadium. The DJfs input, whether msic or speech, is autoaMticdlly overridden when the police control panel is used. Thus a message f m the police can be relayed to one portion of the cxrxJd whilst the rest hear the DJ. The DJ's booth is underneath the police control roam.

?he location of the laudspeakers in the West stand, inside and outside sectors, are shown in Figure 3. These are controlledbybuttons 1 and 7 respectively. ?he 12 loudspeakers inside the stand are mounted under the front edge of the roof and are alternately directed into the upper stand and down to the terracing. These 1- m TOA TZ 201 units. Outside

the stand TDA TC 1% horn laudspedkers have been used. Using TDA data for the lcudspeakers calculations were made of the SOUXI pressure level (SPL) at various locations inside and outside the gnxlnd (see Table 1).

No useful infomtion on typical noise levels in football grounds was found in the technical literature. Consequently limited w e n made at the first available apportunity which happ14 to be the Liverpool vs Arsenal match at Liverpool FCf S Anfield groumi on the 26th of May 1989. Tape recordings were made of the saund l w d at the pitch side and on the collar of a member of RISDfs Noise and Vibration Section who was a spectator in the \Kcpf. The SOUXI pressure lwel was also logged as 1 minute averages using noise dosemeters in both cases. The pitch side micmphones were located appruximately Im frcan the goal line, 8m froan the nearest goal post and 4m frcan the qectators at the \Kopf end of the grocnd. The equipnent used is detailed in Annex 1.

Plots of the 1 minute average sounri pressure level (SPL) are presented in Figure 4. The match kicked off at about 20:15 and finished about 22:OO. llring this period the 1 minute average level in the Kop was always greater than 90 &(A) and above 100 &(A) for much of the t i m e . The corresponding figures at pitch side are a b u t 5 dB lower, viz 85 &(A) and 95 &(A).

Unpublished data fram a f-l match at Bristol, and frcan the Ehrgby mgue Cup Final, at Wembley, were made available to HSE. These data were obtained as 3 minute averages and are presented in Figure 5 together with the HSE data re-calculated as 3 m i n u t e averages. Clearly the HSE Anfield data shows much higher lwels than at Bristol and Wembley.

The logarithmic units used to noise as a 1 or 3 minute average can mask much lmer levels which may have o m x m d during the period. Thus it is possible for the saurd level to be relatively low for a significant proportion of the time whilst the average lwel is high. A statistical distrihrtion analysis of the saund lwels illustrates this effect. Table 2 presents such data for the Anfield recoxdings for 3 minute periods leading up to and just beyond the start of the match. Froan these data it can be seen that although the 3 minute average levels in the Kop were between 97.9 and 105.7 &(A), the level was below 91.9 &(A) for 10% of the total period ( a s *Wn by & and below 96.9 &(A) for 50% of the time ( . Figure 6 ill tes the relationship between the various sta =?Q) lstical -ts of fluchating saunl for a 3 minute period of the Anfield recordings in the Kop. In this period 108.2 &(A) was exceeded for 1% of the time; 105.2 &(A) for 10% and 97.2 dB(A) for 50%; wh- the average lwel (Leq) was 101.1 &(A).

In Figure 6 various crawd activities w h i c h give rise to high levels of sound are identified.

wrdings w e r e made of relayed by the Hillsbow PA system at locations inside and outside the West stand and terracing. The locations a r e s h ~ n i n F i g u r e 7 a n d d m w i n d e t a i l i n A n n e x 2 . T h e r e c o w were nrade on the 7th June 1989 in an empty graund. The speech was input using either the micmphone on the control panel in the police control room or the one in the DJ's booth and consisted of a reading of the team list froan a mat& pmgmme and other lists. Recordings WEE made for two

different voices, one male the other female. The recordings were analysed for their sl-und pressure l w d and anplitude fluctuation. The A-weighted* SFL at each meaflu~ment location is given in Table 1 together with the calculated performme of the PA system. As wuuld be expeckd the meamred saad levels frcm the PA system vary depending on how loudly the passage was spoken, or individual words in the passage. However, it is clear that the system is capable of producing the theoretical lwels calculated froan manufacturer's data inside the ground (Sector 1). Outside the West stand (Sector7) thelwdsmeasuredaresaanewhatlessthanthose calculated.

(* The 'A' fmquency weighting is applied to acaustic signals to sirrmlate the frequency response of the ear.)

The 'Green Guide' (Ref 1) states, in Section 12:

There should be a ammmications system capable of clearly relaying essential messages to all parts of the ground in both normal circurtlstances andinanemeqency."

and in paragraph 194 iii:

ttPublic address equipent should be installed so that broadcast messages can be heard under reasonable conditions (including emeqacies) by all persons of normal hearing in any part of the ground to which the public has access. The system should be capable of being controlled f m a central point should ideally be capable of addressing messages to each major section of the ground. The pawer supply to the system should be such as to enable it to operate in an eaeqency such as fire or failure of the mains supply. It

I undenbrd frcw DP Saund Ltd that the Hillsborough PA system was designed to fulfil this mqubmmt.

Whether a message can be clearly understood will depend on many factors but it was not possible to identify from the literature any criteria by which the intelligibility of speech f m public address systems could be assessed directly. There are standards for emergency evacuation signals and guidance for selecting warning signals but these do not address spoken warnings f m PA systers. Standards for assessing the intelligibility of speech in roams cannat be applied directly to sports stadia PA systems. Therefore it was not possible to assess the performance of the Hillsborough PA system against defined performance criteria in terms of intelligibility of messages relayed throuFpl it.

Typically tests of qee& intelligibility require panels of subjects to listen to standard word lists against qrapriate ba- noises. Such tests were impracticable in this investigation. Instead an ad hoc assessrrent was mde using recordhjs of anmmammts made at Hillsbomugh and of cruwd noise made at Anfield.

In this ad hoc test twelve subjects (HSE staff of age mnge from 21 to 43 years) wexe asked to judge if they could understand the recording fmm Hillsborough when replayed together with the recorded crclwd noise f m Anfield. It was not possible to reproduce the ewct sound field present at Hillsborough at the time of the incident. Therefore the recordings were replayed in an anechoic d.lamber through a PA system which had been set to

produce a uniform diffuse sacnd field a t the listening position (see Figure 8). Ihe two r e c o ~ were superimposed a t krown relative levels andthelisteners-askediftheycouldhearthePA-tsandif they could understand them. The test was repeated a t four relative levels of the two signals. Table 3 lists the level of the -t relative to the CrCkld noise a t which the listener could urd- the -t. Nearly all the listeners could underskmd the -t *en its average level (reg) was 10 dB M m that of the CrCkld noise.

M r Annible informed me that in designing the PA system the installers had made the f o l l ~ a s u q t i o n s in respect of noise lwels a t football matches: anbient levels 77-80 &(A) ; average lwels 85 dB()) ; short duration levels 90-95 &(A). The data fraan Bristol and Wembley wmld support these asscnrpstions but the Anfield data wmld W. Obvimsly the noise level of spectators w i l l depend on many factors such as: rnrmber of qecbtors; design of stadium; signifiamce of the match; emtional state of the cruwd. Each of the cruwd noise -ts presented in this report were taken a t important matches but w i t h very different csrawd numbers and in different stadia. ?his a m t s for the range of noise levels reported. It should be noted that the match a t Anfield was the League Championship deciding match and Liverpool FC fans are noted for their vocal (choral) support for their team. T h e Anfield graund is very ampact w i t h roofing covering all spectator areas. All factors w h i c h wculd lead to a high crawd noise level.

In general a wide range of levels w i l l occur depending on the factors mentioned m e . IXlring any one match or even any period of a match, levels w i l l fluctuate and althcugh the average may be high this w i l l be influenced by short periods of very high lwels , leaving significant periods a t luwer levels (as illustrated in Figure 6) . Messages froan a PA system can be understmd fraan information received in these periods. The simple listening test described i l lustrates this since messages w i t h average sound levels 10 dB (A) belaw the ambient level could be understood. Thus, whilst it is impracticable to design sound systems to W- the highest levels, a cr i ter ia based on exceding a particular level which it is assumed crrckJd noise w i l l only exceed for short periods of time is perhaps more real is t ic (say* There is, of course, a difference in the abili ty to hear ard und a message and actually l k t m b g to it. spectators deeply involved in watchbq the game may not l isten to messages they can clearly hear. A t Anfield, and a t a recently televised International nqly match, spectator safety messages were preceded by a chim which identified t h e m as such. T h i s facil i ty seems a sensible way to at t ract the spectators* attention.

In the particular case of the situation a t Hillsborough leading up to the incident I have no information of a m d noise lwels on the West tenacing. Nor do I have such information for the areas inside arid autside the turnstiles a t w i n g s Iane. Careful examhation of a video tape, -lied by West Midlands Police, for the period 14:45 to 15:07 has not enabled noise levels a t the West terracing to be determined since the \effectst microphone ~ a l s o p i ~ u p ~ n o i s e f r a a n o t h e r p a r t S o f t h e g r o u n d . Ifpto14:50 it is nut possible to identify any chants clearly related to Liverpool fans. Additionally, when spectators are a t the i r most vocal they usually wave their hands, sxrves and banners in the a i r en masse. There is no sign of this activity on this section of the video tape. Tbrcqhcxlt this period nusic is being played on the PA system and can be h- clearly abwe the CrCkldnoise. A t a b c I u t 1 4 : 5 O t h e ~ w e r e a r u x x l n c e d ~ e r t h e P A s y S t e m .

F'mn then on there were the usual chanting of the fans as the i r team members' names w e r e called and as the teams entered the pitch a t 14:54 and the m a t c h started. Eie twem the -ts of the and the kick-of f , msic w a s being played and can be heard except during the loudest chants frcan the crcrwd.

The pattern of crclwd chanting seems much the same as that obsexved a t Anfield th- visual indications frcan the video tape suggest that a significant percentage of the Liverpool fans were not joining in the chanting. ?bus given the open nature of the W e s t terracing and the numbers of fans there it is masonable to conclude that noise levels a t the W e s t terracing would not have been as high as those found a t Anfield. In my opinion it is probable therefore tha t any messages relayed through the PA system would have been intelligible t o the spectators. It should be noted that althmgh the Hillsboruugh system is capable of having a chime, one was not f i t ted a t the time of the incident. I urdersh-d frcan M r I A Rarrrsay, Managing D i r e c t o r of DP Samds Ltd that a chime has nclw been f i t ted following the pblication of the Interim Report by Lord Justice Taylor. (=f 2)

F'rm this l i m i t e d hvestigation I have made the folluwing conclusions:

7.1 ?he Hillsbom PA system had been professionally installed using good quality c a p n e n t s and meets the performance cr i ter ia to w h i c h it was designed.

7.2 It is probable that spectators in the W e s t terracing and its approaches could have heard any messages relayed through the PA system during the period enccanpassing the incident.

7.3 There is limited information on crcrwd noise in sports stadia and the necessary performance -ts for PA systems to ensure intelligible camnuxication.

1. Guide to Safety a t Sports Grounds (1986) HMSO ISBN 0 11 340840 4

2. T h e Hillsbormgh Stadium D i s a s t e r 15 April 1989 Inquiry by the R t Hon Lord Justice Taylor - Interim Report Rugust 1989 ISBN 0 10 1076525

hsq (dBfA)) meamred inside and outside the West stand a t the Hillsboraah smund d u r h broadcasts from the PA system

?he measurement positions quoted are sham in Figure 7 and described in full in Fppendix 2 .

l?mxasT SPL AT m EOSITION

Assumirrg cperation at rated power

Male voice speaking normally on DJ's system. Volume control set to the first dot.

As above but with the speaker using a raised voice.

Female voice speaking normally on DJ's system. Volume control set to the first dot.

~s abare W with speaker using a raised voice.

Male voice speaking normally on the police system to the inside of the West stand only.

Male voice speaking normally on the police system to both the inside and cutside of the West stand.

Male voice speaking normally on the police system to all areas of the ground.

Male voice speaking normally on the police system to the area outside the West stand.

Male voice normally on the police system to th=e of the West stand.

Femle voice speaking normally on DJ's system. Volume control set to the first dot.

As above but with the volume control set to ln3ximum.

Femle voice speaking nomlly on the police system to all areas of the ground.

1 Calculated levels assume: - spherical propagation fmm a point scurce;

- non-reverbrant field. 2 Measured levels are

taken f m a time history chart recordkg of a sequence of speecfi. ?he levels quoted are for the loudest pssages excluding pauses.

3 Distorted autplt.

4 Very unclear reproduction.

5 lhese -ts are dcgninated by traffic noise fmm the nearby Leppings Lane.

Stat is t ical analysis of part of reco rdinss a t Anfield - 26 m y 1989

HEFORE START OF IWKX

m OF REDBDING PERIOD

- m, dB(A) =l k l 0 $0 '90 $9

(B) Pitch Side

STAKC OF RECORDING sarJND m t *(A) PERIOD LI L l o 5 0 L90 L99

AFTER m OF MATCH

(C) Pitch Side

m OF RECORDING PERIOD =l

=l is the SPL exceeded for 1% of the t i m e =l0 is the SPL exceeded for 10% of the time %0 is the SPL exceeded for 50% of the t ime LgO is the SPL exceeded for 90% of the time Lgg is the SPL exceeded for 99% of the time Teq is equivalent steady level wer the period (average level) .

Mts of listenina tests - Relative level (m) of -t from the PA system to the level of the ba- cmwd noise necessarv for intellisibility.

N.33 RFlIATIVE LEVEL OF A N N C U N m FOR I N T E X U G I B I ~ (B) (dB)

Figure 1 Control unit for PA system

I n s i d e Outside

West S t a n d 1 7

Nor-th S t a n d 7 *- 8

l, 0 p .A 9

S o u t h S t - a n d 4 3 (j

Figure 2 PA eyetem sectors

4 Sector 7

0 Sector l - down (Mounted on the underside of the

Sector 1 - into stand front of the roof) ,

Figure 3 Location of loudspeakers in sectors l and 7

Anf ield Liverpool vs Arsenal 26th Mcy 1989

120 -

I n t h e Kop

18:OO 19:OO 20:OO 21:OO 2 2 : O O

Time (hours:mins)

Anfield Liverpool vs Arsenal 26th Mqr 1989

A t p i t c h s i d e

7 0

18:OO 19:OO 2 0 : O O 21 :DO

Time (hours:mins)

Figure 4 Plots of 1 min. average SPL at Anfield 26" May 1989

Anfield Liverpool vs Arsenal Bristol City vs Sheffield United In trod, ISM Yy 19b9

Wembley Stadium, Rugby L e a v e C u p Final

Figure 5 Plots of 3 min. average SPL

_ _ _ - _ - _ _ _ _ _ _ _ _ _ _ _ _ _ - - - _ - - _ - - - - - - - - - -- _ _ _ _ - - _ _ - C - _ - - _ _ - _ _ - - - - - - -

-

TIME

t-- SOUND PRESSURE LEVEL d B ( A )

X - L A - 108.2 d B ( 6 ) SPL exceeded f o r 1% o f t h e t i m e

Y - L,,, - 105.2 d B ( A j . . . . 10% . . . . Z - L?.,:, - 9 7 . 2 d B ( A ) . . 50% . . . . . . 0 - Leq - lC11.1 d R ! G j kverage l e v e l ove r t he p e r i o d

Figure 6 Time-history of noise in the crowd at Anfield 26'. May 1989 (3 min. period)

I

Figure 7 Recording positions at Hil lsborough 71h June 1989

bject

- .. Attenuator

Hillsborough PA recording

Mixer / Amplifier

Anfield crowd recording

Figure 8 Set-up for listening tests

Wrmen t used for recordincrs of Hillsboromh mblic address svstem made on 7 June 1989

Tape recorder rnnnber 1 (Blue) - Nagra IVSJ s/n 4994

channel 1: W 4134 1/2 inch mimqhone s/n 736121 B&K 2619 m i m o n e p-lifier s/n 462127 QSJP adaptor s/n 1200

olannel 2: W 4134 1/2 inch microphone s/n 928525 B&. 2619 miaqhone preamplifier s/n 360352 m adaptor s/n 340

Tape recorder rnnnber 2 (Black) - Nagra IVSJ s/n 1735

Channel 1: W 4134 1/2 inch microphone s/n 666962 B&K 2619 miaqhone prearrp?lifier s/n 702300 CrsJP adaptor s/n 1210

channel 2: B&K 4134 1/2 inch microphone s/n 558290 B&K 2619 microphone preamplifier s/n 702305 m adaptor s/n 149

Additional euuiment used at Anfield on 26 Mav 1989

Sony wal)aMn professional cassette recorder lxlpont Mk I11 logging d-ters s/ns 21639

21632 21627 15997

Positions of recordincrs made at H i l l s b o ~ on 7 June 1989

The micrqhone positions given in this table are shown on Figure 7 and described in full belaw.

Position A At the front of the pen, 0.75m behind gate rnrmber 3 to the pitch, adjacerrt to the step before the gate. Height above immediate gruund level 1.5m.

Position B At the back of the pen, 0.38111 behind the side gate to the Sauth West pen, under the werhang of West terrace seating, 0.5% fraw the rear wall above the second step from the wall. Height above the step 1.5m.

Position C Wide the tunnel to the centre pens, 10.5m froan the end of the a&ended walls at the terrace end of the tunnel. Positioned equidistant froan the side walls of the tunnel. Height above ground level 1.8m.

Position D Adjacent to the end of the actended walls frcxn the access tunnel 0.2m across from the second upright post of the dividing barrier between the tm centre pens. Height above immediate ground level 1.5m.

Position E In line with, and 0.5m f m the end of the blue handrail between the entrance to the West term- standing turnstiles and the access mad to the South stand. Height above grcrund level 1.5m.

Position F In line with the grey sheeted fence separating the Leppings Lane entrances to the West and North stand seats from the entmna=s to South Stand and thewest stand standingarea. 10.2~1 f m t h e end of the fence outside the gates by wings Lane. Height above ground level 1.5m.

Position G At the entrance to the tunnel to the centre pens equidistant fmm the side walls. Height above ground level 1.5m.

Position H In the area inside the turnstiles A-G, 4.5m from gate C and 2.m from the wall of the turnstiles. Height above ground 1.5m.

Position I In the West stand seats; Row 11 next to the blue rail at seat 99. Height 1.m.

Position J In the West s h d seats; Row 28 next to the blue rail at seat 99. Height 1.3m.