Document
N94- 25106
Aircraft Digital Flight Control Technical Review Otha B. Davenport, Technical Director Directorate of Engineering & Technical Management Headquarters Air Force Materiel Command Wright-Patterson AFB OH David B. L_ggett Flying Qualities Section Wright Laboratory Wright-Patterson AFB OH Introduction ..
The Aircraft Digital Flight Control Technical Review was initiated by two pilot induced oscillation (PIO) incidents in the spring and summer of 1992. Maj Gen Franklin (PEO) wondered why the Air Force development process for digital flight control systems was not preventing PIO problems.
Consequently, a technical review team was formed to examine the development process and determine why PIO problems continued to occur. The team was also to identify the "best practices" used in the various programs they looked at.
The charter of the team was to focus on the PIO problem, assess the current development process, and document the "best practices'. A multi-agency, multi-disciplinary team was established with members from Air Force Material Command/Engineering (AFMC/EN), Wright Laboratory/Flight Dynamics Directorate (WL/FIG), Aeronautical Systems Center/Engineering (ASC/EN) (both engineers and managers were represented), and Air Force Flight Test Center (AFFTC) (both engineers and pilots were represented). The team conducted the review in July and August of 1992 and prepared the final report and briefing for Gen Yates, the AFMC commander, in August and September 1992.
The team reviewed all major USAF aircraft programs with digital flight controls, specifically, the F-15E, F-16C/D, F-22, F-I 11, C-17, and B-2. The team interviewed contractor, System Program Office (SPO), and Combined Test Force (CTF) personnel on these programs. The team also went to NAS Patuxent River to interview USN personnel about the F/A-18 program. The team also reviewed experimental USAF and NASA systems with digital flight control systems: the X-29, X-31, F-15 STOL and Maneuver Technology Demonstrator (SMTD), and the Variable In-Flight Stability Test Aircraft (VISTA). The team also discussed the problem with other experts in the field, including Ralph Smith and personnel from Calspan. The following are the major conclusions and recommendations of that review.
Findings: Digital Mechanization First of all, a review of aircraft that have experienced PlO problems in the past indicates that PIO is not a problem caused by _ mechanization per se. PlOs have been encountered with all kinds of control system mechanizations. Mechanical, hydromechanical, electromechanical, and analog electronic systems have all encountered PIOs in the past. Table 1, from Reference 1, shows several PlO problems that have occurred in the past.
However, digital electronic flight control systems have allowed us to break the space, weight, PRlfP.,,ED_£; PAGE BLANK NOT FILMED and power barriers that effectively limited the flight control complexity that could be achieved with other control system mechanizations. With digital flight mechanization we can tailor the flight control system for a far wider variety of flight conditions and flight tasks than was possible before. This added complexity adds some additional risk that may require a more disciplined, more structured process to manage in the development process.
Findings: Development Process All of the programs we looked at used pretty much the same development process. A simple schematic of that process is shown in Figure L This process is inherently iterative. Each step is intended to better identify the system and reevaluate the system based on the latest identification.
When problems are encountered the design should be modified, re-identified and reevaluated, when problems are overcome the process moves on to the next step. This process is intended to reduce risk as the uncertainty decreases. Our conclusions about the process were that the process had the right steps, but the execution varied from program to program.
In some programs, the twin constraints of cost and schedule sometimes drove the process to run "open-loop" when flying qualities problems (including possible PIO problems) were encountered.
For example, if a design did not meet the quantitative requirements in the specification and the necessary fix significantly impacted cost or schedule, some programs discounted the applicability of the requirements and decided to proceed with simulation to see if the problems existed. If problems were encountered in simulation and the necessary fix significantly impacted cost or schedule, some programs discounted the fidelity of the simulator and decided to proceed with flight test to see if the problems existed.
Findings: PlO Figure 2 shows a simplified schematic of the pilot-vehicle system. The pilot can be viewed as a feedback system that closes the outer-loop around the airframe-sensor-flight control system. The feedback path for the pilot is a multi-channel path that includes the pilot's visual cues (outside and inside the cockpit), motion cues, aural cues, tactile cues (force and displacement) from his controllers, and others. A PIO occurs when this outer loop becomes dynamically unstable or neutrally stable. In the most general sense, a PlO is the result of a disharmony between the pilot's action and the expected aircraft reaction. This occurs when one or more of these feedback cues provide confusing or even con.flicting information to the pilot and his gain is high enough to drive the outer-loop system unstable.
PlO susceptibility is when the aircraft possesses certain characteristics that make it prone to get into a PlO in flight conditions and tasks in which it must frequently fly. The typical causes of PlO susceptibility are well known: high stick sensitivity, excessive system phase lag, large system nonlinearities, lightly damped response modes, unstable response modes, coupled response modes, etc.
Each of these problems causes some kind of disharmony in one or more of the pilot's feedback channels.
However, the presence of such characteristics does not mean that the aircraft will PIO all the time. There are other factors involved as well. First of all, a PlO is more likely to occur when the pilot is performing a "high gain" task, that is, he is trying hard to minimize an error in aircraft attitude or rate. Such "high gain" tasks include precision landing, carrier landing, aerial refuelling (particularly probe-and-drogue), LAPES, close formation flying, target tracking, etc. A PIO is more likely to occur in these kinds of tasks than in tasks where the pilot is only loosely monitoring aircraft attitude or rate and making occasional corrections.
Thepilotis afactor in theprobability of aPIOoccurence because a pilot can learn to avoid PlOs in a specific airplane by learning the tasks and the conditions in which that airplane is PIO prone, and learning to avoid it by lowering his gain in those tasks and conditions. Thus a PIO is less likely with a pilot who is experienced with the airplane's PIO tendencies and has learned the appropriate technique to avoid it. PIO is more likely with a pilot who is unfamiliar with the airplane or is unaware of its PIO tendencies. The fact that the pilot is a factor in a PIO should not be interpreted to mean that the pilot is at fault. PIO susceptibility is a design flaw because the aircraft is supposed to be designed such that a pilot can command the necessary degree of precision to do the task without fear of driving the outer loop unstable. An aircraft can and should be designed such that it is not PIO prone in tasks or conditions in which it must commonly operate. The team struggled with the perception that such a design might be impractical from a cost, weight or performance perspective until a very high performance front-line fighter was considered that had never had a PlO and was clearly in the "good" handling qualities regime. This aircralt had set the standard in cost, weight and performance. It was not designed specifically for PIO but careful attention had been paid during it's design to the characteristics that cause good handling qualities.
Sometimes a PIO is initiated by a discrete event, commonly called a "trigger event'. A trigger event is not necessary for a PIO to occur, nor will the identical trigger event initiate a PlO every time.
This is because the trigger event is not the cause of a PlO, it is only a catalyst. A trigger event could be something related to the aircraft such as a discontinuity in the control system (e.g. a sudden failure or a large discontinuity in the control law gain schedule), or it could be something totally unrelated to the aircraft such as a large, abrupt atmosperic disturbance or a pilot distraction. In a PIO prone aircraft, the trigger event will initiate the PlO by causing the pilot to make abrupt corrections, and the PlO tendencies (due to whatever factors) will provide the "confusion" that sustains the PlO, If the aircraft is not PlO prone to begin with, the trigger event will probably not cause a PlO because the pilot can apply sudden corrections without becoming "confused'.
Of all of these factors, only the aircraft susceptibility and certain trigger events are within the control of the designer. Mission requirements may demand that certain "high gain" tasks be done.
The aircraft will be flown by pilots with a wide range of experience (the only way to gain experience with an aircraft is to start learning without any). Certain trigger events are random events with a high probability that they will happen to someone sometime in the aircraft's service life. In order to design an aircrait that is not PlO prone the designer must control those well-known factors that cause PIO susceptibility. The difficult question for the designer is "What values of these factors provide the appropriate level of PIO resistance?"
The reason that this is a difficult question is that, like all sciences that involve the human element, flying qualities issues, including PIO susceptibility, have the characteristics of a "soft" science. That is, since a human being's appraisal is the measure of merit, it is very subjective in nature, and highly variable depending on what human being is doing the evaluating. This variability exists in both the research end, where you are trying to develop criteria to address the problem, and on the verification end, where you are trying to prove that your delivered product is satisfactory. Thus, there is not necessarily an absolute answer, but instead a certain probability based on evaluation by a number of human beings.
The nature of the problem is illustrated in Figure 3. Cooper-Harper pilot ratings are the most common quantitative measure used in flying qualities evaluations. For a typical handling qualities experiment, the correlation curve of a parameter that correlated with Cooper-Harper ratings would typically look something like that shown in Figure 3. At the "good" end of the curve, there is a certain point up to which, in a typical experiment, all of the pilots will agree that the aircraft is good, and the diversity of Cooper-Harper ratings will be small. At the "bad" end of the curve, there is a certain point beyond which all of the pilots will agree that the aircraft is bad, and the diversity of Cooper-Harper ratings will be small. Between these two points is an area where it is more difficult to say precisely how bad the aircraft is because the diversity of pilot ratings will be much greater at any point in this region than at the ends (References 2 and 3). Consequently, the objective, open-loop requirements derived from handling qualities research must be considered inferential in nature. That is, meeting them will provide a high probability of having good handling qualities, but it does not guarantee good handling qualities.
Findings: Flying qualities specifications The quantitative PlO criteria available in the currant flying qualities specification, M_-STD- 1797, and from other sources, are based largely on data generated in experiments conducted on ground-based and in-flight simulators in the 1960s and 1970s. The review team found that of all the available criteria, no one criteria seems to be universally accepted by the community at large. In the flying qualities specifications, most of the quantitative PlO requirements resided in paragraphs that were intended to assure good overall flying qualities, not just to preclude PIO. For example, in MIL- STD-1797 requirements on phase lag in the pitch response reside in paragraph 4.2.1.2 Short-term Pitch Response. In the specifications, paragraphs intended explicitly to preclude PIO problems have been largely qualitative in nature ('there shall be no tendency for PIO'). Finally, the verification requirements in MIL-STD-1797 do not specifically call for testing for PIO characteristics. The lack of a strong tie between the requirements and the verification at each stage of the process has led some programs to defer critical actions at a time when small changes could have precluded a much more significant change later on.
Findings: Flight Test Phase The final test of the flying qualities and the PlO tendencies of an aircraft is in the flight test phase. The problem with waiting until the flight test phase to determine the degree of PIO susceptibility is that by this point in the development the number of realistic options to resolve problems is dramatically reduced, and design changes at this stage have a greater impact on cost and schedule than at earlier stages. Often a cheaper and easier solution at this stage is to train the pilots to avoid the PlO if they can. Consequently, a system with a PlO tendency sometimes does not get fixed unless the pilots cannot find a technique to avoid the PIO or it prevents mission accomplishment.
Conclusions As a result of these findings, the Review Team concluded that the process, as currently implemented, had the the following flaws: .......
1) The available criteria and analysis methods are inferential in nature, they lack universal acceptance, and the current test techniques are not rigorous for PlO problems.
2) Because of this, the current process lacks firm go/no-go criteria at each step in the process for the manager to assess the risk of PlO and decide whether to proceed or whether further iteration is necessary.
2 46 3) Consequently, withregard toflyingqualities ingeneral and PIOin particular, the development process tends to be driven "open-loop" instead of as an iterative process.
4) Finally, the decision of what is good enough is typically left until the flight test phase, where many options that were available in previous development phases are now precluded by cost and schedule constraints, and changes are made only if the pilots cannot be trained to avoid the PIO or the task cannot be modified and retain it's military utility.
Recommendations The Review Team made the following recommendations to resolve these problems in the process.
First, establish an Integrity Approach for flight control similar in nature to those established for structures and propulsion. The intent of this program would be to change the paradigm from one of "proceed unless a PIO problem is proven to exist" to one of "proceed only when a PIO problem is proven not to exist'. This would be done through establishment of firm go/no-go criteria for each step in the development process. At the design stage it would consist of improved flying qualities criteria.
However, since these would still be inferential in nature, further "gates" would be established at other steps in the process. Rigorous demonstration maneuvers, such as Handling Qualities During Tracking (HQDT) would be required in early stages of the development process, such as ground simulation. In- flight simulation would be recommended, perhaps even required if results were inconclusive in the earlier stages. Finally, the verification of adequate PIO resistance would not just be compliance with the inferential requirements, but also satisfactory handling qualities in the demonstration maneuvers during flight test. With the requirements and verification agreed to between the Air Force and the contractor, this process provides a relevant measure of the capability of the aircraft to be operated by the vast majority of the pilot corps.
The second recommendation was to establish a Flying Qualities Working Group in each SPO that has an aircraft under development. The initial purpose of this group is to conduct an assessement of the system and attempt to achieve the appropriate balance between design, pilot-training and military utility. This working group consists of engineers from the SPO, the contractor, the laboratories, and the Flight Test Center, and the test pilots from the contractor and the Flight Test Center. The purpose of the Flying Qualities Working Group is to monitor the progress of the flying qualities of the design through the development stages, help resolve problems, and insure that potential problems are communicated to all the agencies involved.
The third recommendation was to enhance the flying qualities research program to improve the criteria and analysis methods available. The objective is to resolve the conflicts between existing criteria, develop a more comprehensive analysis method, and, hopefully, reduce the region of uncertainty in the present predictive methodology. Another objective would be to develop criteria and analysis methods for new flight regimes (such as high angle of attack) and unconventional response modes (such as direct lift).
The fourth recommendation was to incorporate the "Best Practices" into a new tool being developed for the SPO engineer called the Air Force Acquisition Model (AFAM). The Review Team identified 22 "Best Practices'. Space limitations preclude listing all of them here, but they are summarized below: 1) In the requirements definition stage, use quantitative PIO requirements in the specifications, with specific verification requirements.
2) In the design stage, use multiple analysis methods and criteria to assess the flying qualities of the design.
3) Keep the needs of flight test in mind during the design. For example, include a means to change control system gains during the flight test phase in anticipation of the need to adjust them in order to resolve problems.
4) Ground test with hardware in the loop to identify system characteristics.
Use fuU-up ground simulation and in-flight simulation to assess handling qualities and PIO tendencies and use well-defmed "high gain" pilot-in-the-loop tasks.
6) In the flight test stage, use well-defined "high gain" pilot-in-the-loop handling qualities testing (HQDT, etc.) as part of the envelope expansion process.
On 5 Feb 1993, the findings and recommendations of the Review Team were briefed to the Commander of Air Force Materiel Command. He has directed that AFMC implement the recommendations.
As a result of these and previous briefings to the senior leadership of the Air Force, the "best practices" are being included in the AFAM for use in current and future Air Force programs.
The SPO's either have or are now forming the working groups and conducting assessments to be reviewed by the Program Director. The Air Force Science and Technology program funding for flying qualities has been increased by over 100%. Finally, the Commander of Aeronautical Systems Center through the Directorate of Engineering is planning to release a draR Integrity Program standard by the end of 1993. The focus of the Air Force on the total system requirements for affordable, capable and sustainable aircraR that meet the users needs has been improved by the contributions of all of the team members.
References 1. Ashkenas, Irving L., Henry R. Jex, and Duane T. McRuer, Pilot-Induced Oscillations: Their Cause _, Norair Report NOR-64.143 and STI Report TR-239-2, 20 Jun 1964.
2. Wilson, David J. and David R. Riley, Cooper-Harper Pilot Rating Vic'iabiljty, ALAA Paper 89- 3358, 14-16 Aug 1989.
3. Riley, David R. and David J. Wilson, More on Cooper-HarDer Pilot Rating Variability, AIAA Paper 90-2822, 20-22 Aug 1990.
4. Military. Standard. Flying Oualities of Piloted Ail'¢r_fl, MIL-STD-1797A, 30 Jan 1990.
TABLE I. Some Past PIO Problems (Taken from NOR-64-143) Examples sho'iwl llS: SpIE_ll__q (Alrr_hl: _,*tesl Sul'_l@m; _._,*h-ll Eli*hi Cnm, tltia_s : l_mJirlral CLASS 1 L I II. Illull IIOHIJ_IAII I[l_dm II. SUISlDi41y _lm_IACI NONUNIAII IMPROI_Jt ZlMLII.ATI._IN: 11_ Y_ i-* PORP("JI_ll_ICJ I*Kll2_.l_: p*_ e_ Xystlrlshl in IIkk ckrvator _llml _ kl low bequmKy ^bmNrmalh li_ _ el irr_aml sl_ lud ¢Umk *_lll*_kl. ov_comena _ frlctlms lind rtd-_ces ap_emm 441m_ OI IIw Ihlllll hi r0sp4,1m0 to |Iraqi .L __ UAI*/I_PI,'_P_ (I'.Ai_.D. Ir-lae_",_ ITS.. JL" Ilqllll. IISUII_ II IIIlll ol¢lnllllllll Ii dM_ Ir_J_.I1NIIMJlL"I_D P_ltleli_lEI (i_=l_* V*i,_ I,kd_ pi cm.wlS_ _lllll _ li n_ _._ b: Ij I hun rUll,-_k_mll mm I_1, mil_ml ,_ dlm iraqi.
le _1¢_ c_smcml kd ks mmxbk -IJ_ (]{1_.1_1_ Ir.101R_ I1.1_L4.1; V; I: Uns_le IrJnk in M(o) cm_ Id w mod_ra_-ImSd AIUUI fin S"rl(_[ IA41_|_ T-IlL&I: F: A: Ann mus lecl_*_ led rlmm hmU_; leeds _ B reedbstk mutst4* ¢m_ w4m a_-pefle4 .... Clmml-loep mid _ Um• llq, hl_. Ik_llO_ lrlll_ll_ll{_ / illlOlllfilll IHi¢l_ll_l_l I: _ _1 _IL'_,,_ ;;_J_ll_ liMd_ i_ I_lm dtl ,'-*-L Ni Irmiilm_.
I%,;Udl • L0, k*_S m cheed-l,_ • . o o n y.- L- Spg_l _1 m,lk iIl_lmu- ml, k II _11_r_m_si_ b dq_kd 411H_I imml_.
TLAN_ONII_ _ tA_DI--- Y_ FIJFL _ SNAI[L*40 (I(_-L_A_ T-_TAI-* Y: £.* _ llllh Sep*_mm ov_ m4du cumin ¢oM_ _ rw yAW ¢m_p/*s wire dmcb _11 mede g,lml mini Io B YIiW I)l¢t_l|kllll. le dims _llW _ll_ll01m.
HONE KNOWN IIOU. Slmdl mmll ¢7¢il dill Io dalnpmr lilplaqmtNI Illtmpilld Io ¢OClIIol II.
"CHd¢_/kk ml_ I. I,**Ikll m.m_il **_ llmm I • lift, Ilk "4 • VIlli (sk/_mm) e • Cmi_ [ specs,
i sT°s I
? _i_ IsYNT"ss's H 1
" _._ _"",,,-,..J '_' --'---][---I_IGROUNO /
u
I
_ c _ _ & &I_;-T .... H EUG.T
S T y ACCEPTABLE FIGURE 1. Simple Schematic of the Development Process FIGURE 2. Simplified Schematic of Pilot-Vehicle System
PILOT FCS NC
Task
Inner Loop
SENSORS
Outer Loop
FIGURE 3. Pilot Ratin8 Correlation With a Flying Qualities Parameter S Worst pilot rating Uncertain ¢D ........ ,-- Average pilot rating c- Best pilot rating Good Bad 4-J O m im
D_
• - - o
Flying Qualities Parameter
25O
(/)
:3
LL
37_
I ....... II III II I I I II b_,., 5-.
_o_.=
I" =-- -,,, PRECEDING PAGE BLANK NOT FILMED I gll I I Ill|Hill
@
>-
r_
LL
!--
(/)
Or)
i
Z
m
(/)
Z
LU
n- I
..J
m rr ..j u- E: Lu _ z
> u) 0
0 uJ o z _
I-- E LU 0
I
o :E (/) 3
@
i!1 --- :3 --J
E 0 0
0 _u =: z
, 0
EC:
O__ m
im _3 O_-- ..C r, i.i L.
O.._>
¢- i-.
a._3 m gB IR
_'o
O'-
v-a._
........... I IIIII
(,3
_2
,<£
N
I:=111_i
B:!
_'
C]
I [ I CO {0!
::) UJ
©
U.
I,L L! .....
Z Oi
[ J
< > LL Im i -- 0 •-- _ I
<
>..
0 "0 >, I tn "0 4)
0 E
r_ • | Im ... -- O. I
0 _: 0
- _ o 0 I
m O , , _ ' _I" i ' CO _ i _ I
<
I- ,,.::.: ........ .z _........... ._ K,._.._....# _.._.__,._
W
;<,:r:,:"
C_
<if
_: E= Gi
E _9 QE
__ LI,t c ' E 0 ,- 0 lc:l lc:l _1 Ic:l
_o
._o ®
.__. c'o
o°
:3 13) 0"0
.c _c
._or) '113"0 =3 U.
__o__
IIIII Ill I (/) uJ Z <
0 E
_J m rr
@
m < < D.
m < (I) iii m rr W iii
C)
I_ .,m m D: 0) EL I%: rr E-- Z 0 IJ- ill IL U)
=E
..I !11 < E F-
E c
Z tw_.m Iii ILl E (/I J i i ILl ,1-,, E .:" O_ 0 (/) < U.
<
0 _) 0
ILl )- Z I- • ._ m F- m 0 < n F-- F- ro .< )-.
a_ • • • Z _J .< <{ >
0 iii
eu, II III II L _ C m 4=-=
.__ •
_.__
.__ _
"i
_ .__
.__._ _ ,_
I III I I
I._I
C_
LL
oE_z s°-
.__ _o .m •
._ _8 _-
_._ _-"
® ._ ._m
L> e
I I I IIIIII II LU D_
0 0
_J LU
@
ILl U} a UJ
U_
<
3 gy
<
lm
<
0 "--
f'_ • •
L_B • • _ • • • 0
_i 0
j.-
0 .-J
Z
0 < mm_o
I:I I_I I I
-o_
'U
__>_
I--_m I--- ...
m X a: U UJ
w _ z _i
-- !- 0
!-
Z.
Z
@
I-
I+
IL
rt Z I=1 iii IIZ W Z
0 0 Im
u n _i i_ 0 IX, t_-- n 1:7: Z W 1:3 0 IX: ..1
I
I- 0 Z
C+
N o
!&:: o
Ill
m 0
• • • .-I Z W t:::_+:: E O) nnn _mnmm I_ I11 II II t I
oo___
U)
._.._- .-- o _ o _
_- _- o _._.__
©
llllllll m ,,J
<
Z Z
_o
I- .J ,¢ Z :3 m
<
E: n LU Z n _l v
!Z
, (#'l $ Z < m Z 0 :::) iilel_ ' I- Z m
< :i
<
i:!.! :! 0
t
i,U + LI.
0 x
Li,,I > ::) I#l I- ri- ttl _- fl,"
z ,<
ill Ei !-- I--
,171, _ i_ _"
I- I:1 1_1 1:1 II
__=
E-a
-a 8 -=
._o _==
O.__ =
_.__
8_
I.-= ==
,-=._ "=
_= _._
_._ _.
I I I IIII I
c_
:)
(,9 V
C9
=8
CC
(2:
LLI
LLI 0 _ (:},.
l,.- f_ ¢o (/) I-..
I..- I-.
h. i mli Ill 11111 e_ c _ n ._ 0
,_.-_ ,__
m m
o_a.m
__o_
o _ "_
" m _ C 0 e-
•-_ o_.$
E
i II III II IIIII Iv
t'J
u.l
I.U CO ill inlll IZl m
I--
I11 .... I I_ +t_+,:,; I
<
Z_ ..>+1 m l.l., Z
)
. N.
I+,, Ill Z_ I
z
m
> 0
<
m _.
o <
......... J_ I]
*-°8
=._ ,.
a.
°_. 8! 8._°
_._ _ g
lilt IIIIIIIIII
C)
CO
©
i........
[!.!
C_
t tttt III r
<
0 £L
<
<
,=,. (3
<
LU
_Ln
I
I.._,.!
..I I_I II I L 276 . .
C _-£.
i
o
°
m_Q,.
C m
•-_ I--- m
t_ III IIII III
"6
Z
C_
l[
¢)
I=
L,t,_]
L_
2"/8
o m
o_
.c.- C
._ o
0) 0 _) :"
E_
Q.
u) C'O 0 C (U (U £1.
m wmm ,m 0 0 ,-
8_
I II I IIIIII II ..J X UJ _D T-- | 14.
UJ .C: LI.
ii -- ffl Z m C Ii ffl ! i I II II
fZ_
L:;:,.
_Nll lIT ORt_NAL PAGE 1t5 OF POOR QUALITY O_
8_
m t_ _d
-_.q
o._ _
n_ 'It"* 0 _ _.
:E
Z) >-
0:: d*
LU 2: I Ill LLI
gNmi_
m I .@ ILl Z m C_ NN_ Z W LU .J f_ N@ Z N N: f,n N _xi I:::L '_ 0"0 ,._._ ®I::
_.c_.._m
,-.0 ,_ o1::(1) mm
_e.[
.-_92
..C C. C I-" 0 .,- I I I II e-- >.
C)
W Z .J LU cO <I:
(.._..)
LTI !
,#6:" "_ <
.L)
tll W Z Z tU Ul J Z ii i ili Q • e_ W 0.1=
o°.o
.=-
|J im m £: £: QlU Q.
01 G >., l£J >-
o_+o
O
-+_O _j
I++..
O m ,,_
o
o _ F_.
©
II.I L_ t_r,) I_
i-ua_ o
O_m u.
i+++ LLI
LLi I+lJ _1 I++" LL!
<o "
L..t. .... Z m
3m o°
ILl W _1, Z l.l.l LU I+L,_) l-- CO I +- 1_ C_
o
__.+o ,+u <
o _
Z 030 D m 13.
::'+:_ O
0 _: u,,l
I_N
0 m 1_
" 2: tu _: ::N::.NN
0o_ 0
+_ _:'+::'+::+ O _.':::" ::: ammm .... _|+ LL.
.. _-O+J
_N m
0 0<_ m
LL! O
z Nlu_
0 +uJ_ _ ....
-- , __.,_(J) :30--+) +++.:.:} UJ t _+ LL! ltJ 0'} Or) i:::)
LL+
11.
I".
_- • • • • • • • • • n LI.J nnl I:t 1_ II II III _ °
_o o_
• _,_ _ .___
0 _ _.c: I I II IIII III I O)
.__ '_
C)
<
f_
L!J
¢_
03 iil
0 ::)
-J -1- F-
, :::)
< z u.
III I