The 1st International Conference on Information Science and Engineering (ICISE2009)

Dynamic Risk Assessment for Excavation Engineering Based on Human Factors

Jian ZHOU 1,2, Qunyan TANG 1, Jiao Zhang 3
1

Department of Geotechnical Engineering, 2Key Laboratory of Geotechnical and Underground Engineering of Ministry of Education, Tongji University Shanhai, China

Department of Civil Engineering and management Shanhai, China

Shanghai Technical College of Urban Management

AbstractThe excavation of foundation pits is usually confronted with many potential risks during the long construction time. Risks often come from varies of working procedures. Many factors may cause the procedures failing, but human errors show the greatest probability to cause accidents. Luke Model is a valid method of quantitative analysis for human factors. Through quantitative analysis, get the failure probability of excavation engineering caused by human factors. Based on these probabilities, use method of entropy weight measurement to calculate risks degree and coefficient of entropy weight of each human factor. Finally control the risks of each working procedure according to these two indexes and eventually get the aim of reducing risks maximally. Keywords-Risk; Dynamic Risk analysis; Human factors; Probability; Consequences

Figure1.

Process of Risk Analysis

Failure condition probability of each procedures based on human factor (Qki )

Hazard identification based on human factors

Failure probability calcultion

Failure probability calculation for each procedure in excavation engineering Identify procedures with larger failure probability (totally n kinds)

I.

INTRODUCTION
Expert investigation chart, engineering experience and accidents statistic

As a complex systematic project with big investment, the excavation of foundation pits is usually confronted with many potential risks. While human errors are the greatest probability to cause the risk accidents [1]. The concept of human factors has first been put forward by Chapanis 1956 [2]. And Abdelhamid (2002) used it in construction project [3]. Vicknayson (2004) emphasized that risk management must consider human factors [4]. In 2004 Cacciabue P.C. put forward risks management theory of human errors, provided a valid method for human errors risk analysis [5]. But according to our best knowledge, there is no systematic study about human errors during the excavation of foundation pit to date. It is very interesting to model the occurring and the influence of human errors during the excavation and build a defense system to various potential risks. II. THE PRACTICAL IMPLEMENTATION OF RISK ANALYSIS

Risk analysis with entropy weight method

Risk analysis

Calculate the human factors weight of each procedure Calculate risk degree of each procedure

Control the high weight risk factors according to the different procedures

B. Calculate failure probability with Luke Model There are several mathematical models describing human factors. Luke Model is one of these models and it is suitable for the excavation engineering problem. Its related assum There are several variant working procedures for complete a task;

A. Overview For considering the human factors sufficiently, we used models of human factor to analysis the failure probability of each procedure quantitatively. Based on these results, analyse the risk degree of each procedure for the aim of reducing risks. Fig.1 shows the process.

978-0-7695-3887-7/09/$26.00 2009 IEEE

4270

Each work may be done more than once in task and several kinds of errors may exist in one work procedure; Error modules in work procedure are independent; Only one error, the whole task may fail and also may not. The occurring probability induced by the kth error module of the ith work procedure is as follows: Fki = qkiQki 1 Where qki is the occurring probability of the kth error module of the ith work procedure, Qki is the failure condition probability of the ith work procedure induced by the failure of the kth error module. Then for all M i kinds of error modules, the failure probability of the ith work procedure is as follows:

1) Risk Degree Risk of excavation engineering includes two facts, occurring probability P and its consequence C. Risk degree (R) is likelihood predication of the occurring and consequence. It can be defined as the follows. 7 Where f means error doesnt occur, s means error occurred. 2) Coefficient of Entropy Weight Use method of fuzzy logic to evaluate the probability and consequences. aEstablish Fuzzy Set Construct set of risk factors as U = {u1 , u 2 , , u n } and b) Establish Set of Membership Degree Experts evaluate the factors in set U according to evaluation set V, eventually get the membership degree set as

R = f = 1 Pf C f = 1 (1 Ps )(1 C s ) = Ps + Cs Ps C s

Rnif = 1 (1 qkiQki )
k =1

Mi

evaluation set as V = {v1 , v 2 ,

, v n }.

If the whole task includes

S i times of the ith

3

independent work procedure, then the success probability of the total task is as follows:

Rnti = ( Rni ) S i
Where

S i is the times of the the ith independent work

procedure. When all z kinds of work procedures are not influenced, the total failure probability of the task can be competed by

8 c) Determine Coefficient of Entropy Weight For one risk factor, its relative significance can be measured by value of entropy, which is computed by

r11 r R = 21 rn1

r12 r22 rn 2

r1m r2 m rnm

RF = 1 ( Rnti )
i =1

ei =

Si

1 m rij ln rij ln m j =1
1 (1 ei ) nE

In fact, these z kinds of work procedures often are not independent. If these procedures have great dependence, the total failure probability can be demonstrated as follows

Normalized the above entropy value, can get coefficient of entropy weight.

i =
n

10

RF = 1 max[ (1 qkiQki )]S i

i k =1

Mi

Where E 5

= ei , and i should be satisfied with

i =1

In excavation engineering, these procedures are neither independent nor corresponded completely. So the failure probability can be expressed by

0 i 1 i = 1
i =1

11

1 max[(1 qkiQki)]Si RF 1 (Rni)Si

i k =1 i =1

Mi

C.

Analyse Risks

3) Calculate Ps and Cs Assign A=( 1 , 2 ,..., n ). Assess weight for each factor in evaluation set V, gain weight vector as B=(b1,b2,,bm). Then occurring probability of the risk accident can be calculated by 12 With the similar calculating method, consequences can be showed as

Entropy means a measurement of uncertainty, it was first supplied by Shannon (1948) [6]. Then, E.T. Jaynes (1957) get Maximum Entropy Principle [7], which can describe this kind of uncertainty. Using entropy weight coefficient to measure the uncertain risks in excavation engineering is suitable [8].

Ps = AR p B T
T

C s = ARc B

13

4) Risk assessment After calculated Ps and Cs, , R (risk degree) can be determined by formula (8). In general, when R 0.7 , the

4271

system are front with high risk R 0.3 means low risk and between the two limits, should adopt some measures to control the risks. III. EXAMPLE A. General situation The excavation engineering locates in Shanghai with depth of 10m. Mainly adopt drilling compresol piles as its exteriorprotected construction with rabbling piles for water-sealing. The site mainly used for resident and market. There are many pipelines under it. Surface is plane with ground elevation of 3.563.87m. Geomorphy is singly strand plain type. Since Shanghai belongs to soft soil area, this site is unfavourable lot for quake-proof. B. Human Errors Identification In this engineering there are eight procedures .The procedures and their relative human errors are showed in Table.I.
TABLE I. Procedures (a) Pile driving (b) Ring making beam HUMAN ERRORS EXISTED IN PROCEDURES Probable Human Errors 1) Material deviation 2) Construction Techniques errors 1)Material deviation 2) Shaking less compactly 1) Wrong excavating method (c) Earth excavation 2) Exceeded Excavating 3) Exceeded Load 1) Material deviation 2) Shaking less compactly (d) First poling 3) Incorrect Maintenance Method 4) Axis Deviation of Upright Column 5) Weld Flaw 1)Wrong excavating method (e) Earth excavation secondly 2) Exceeded Excavating 3) Exceeded Load 4) Digging machine Collision to exteriorprotected and poling construction 1)Material deviation 2) Shaking less compactly (f) Second poling 3)Incorrect Maintenance Method 4) Axis Deviation of Upright Column 5) Weld Flaw 1) Wrong excavating method (g) Earth excavation thirdly 2) Exceeded Excavating 3) Exceeded Load 4) digging machine Collision to exteriorprotected and poling construction 1) Material deviation 2) Thickness shortage of Floor (h) Bedding cushion making 3) Shaking less compactly 4) Bottom sealing flaw 5) Pouring delay of bedding cushion and floor

C. Evaluation of human errors probabilities Considering the human errors listed in Table.1, each procedure has several probable human errors. Based on experiences, get the occurring probability and conditions probability of each human error. For procedure (a), there are two kind of probable human errors, material deviation and construction techniques errors. From our investigation, static and experiences, the occurring probabilities of these two errors are 5% and 10%, condition probabilities are 5% and 70%, the failure probability of procedure (a) can be calculated by

Rnbf = 1 (1 qkbQkb ) = 1 0.98505 = 0.01495 14

k =1

Other results are listed in Table..

TABLE II. Proce -dure
(a) (b)

FAILURE PROBABILITY OF PROCEDURE Occuring probability (qki)

5% 10% 10% 20% 10% 10% 10% 5% 10% 10% 10% 5% 10% 10% 10% 10% 5% 10% 10% 10% 5% 10% 10% 10% 20% 30% 5% 5% 10% 10%

Probable Human Errors (k)

1) 2) 1) 2) 1) 2) 3) 1) 2)

Condition Probability (Qki)

5% 70% 5% 5% 5% 5% 5% 5% 20% 5% 20% 20% 15% 20% 20% 10% 5% 40% 5% 40% 40% 20% 30% 25% 20% 5% 5% 40% 10% 30%

Failure Probability (Rnaf)

0.07230 0.01495

(c)

0.01493

(d)

3) 4) 5) 1) 2) 3) 4) 1) 2)

0.03700

(e)

0.09170

(f)

3) 4) 5) 1) 2) 3) 4) 1) 2)

0.10360

(g)

0.11020

(h)

3) 4) 5)

0.07530

4272

iIf these 8 procedures are dependent completely, the task success probability can be computed as follows

TABLE IV. Probability u1 u2 u3 u4 u5 u6 u7 u8 u9 v1 0.4 0.6 0.5 0.3 0.5 0.4 0.4 0.3 0.4

MEMBERSHIP DEGREE SET (RP) v2 0.3 0.3 0.3 0.3 0.2 0.3 0.2 0.3 0.3 v3 0.3 0.1 0.2 0.3 0.2 0.2 0.2 0.3 0.2 v4 0 0 0 0.1 0.1 0.1 0.2 0.1 0.1 v5 0 0 0 0 0 0 0 0 0

RT = Rnti = Rni = 0.5807

i =1 i =1

15

Failure probability is as follows. 16 iiIf these 8 procedures are related completely, the success probability of task can be computed as follows S R T = max (R ni ) i = max (R ni ) = 0 .98507 17
i 8

RF = 1 RT = 1 0.5807 = 0.4193

Failure probability is as follows. (18 iiiFrom the above two probability, the success probability of excavation engineering based on human factors can be determined by 19 0.5807 RT 0.98507 Failure probability is as follows. 20 0.01493 RF 0.4193 Entropy Weight Calculation From the analysis results of human errors with Luke Model, can find the failure probabilities of procedures of (e), (f) and (g) are relevant bigger. Analyse these more critical procedures with method of entropy weight measurement to get the risk controlling key. 1) Procedures (e) and (f) a).Probability Calculation (Ps) Construct fuzzy set U = {u1 , u 2 , , u 9 } and evaluation set V = {v1 , v 2 , , v5 } , where ui refers to the 9 kinds of human errors in procedures (e) and (f) , vi mean as Table.. After investigation from experts, can get the membership degree set as Table.. Calculate ei through formula (9), get entropy value vector as ei = 0.677 0.558 0.64 0.816 0.758 0.795 0.828 0.816 0.795 , and with (10), calculate entropy weight vector as A = (1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 ) = 0.14 0.191 0.1560.0790.1040.0880.0740.0790.088 Assign weight for {v1 , v2 , , v5 } , get B as B = (1 / 25,3 / 25,5 / 25,7 / 25,9 / 25) . Through formula (12), gain Ps as Ps = AR p B T = 0.11
TABLE III. Probability Grad v1 v2 v3 v4 v5 PROBABILITY GRADE DEFINITION Explanation Risk are impossible occurring Risk has little probability to occur. Risk may occur. Risk has large probability to occur. Risk almost is certain to occur.

R F = 1 R T = 1 0 .98507 = 0 .01493

b) Consequences Calculation (Cs) Construct evaluation set V = {v1 , v 2 , , v5 } for consequences, vi mean as the Table.. After investigation from experts, can get the membership degree set as Table.. Calculate entropy weight vector as

D.

A = (1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 )
Through formula (13), gain c) Risk Degree
TABLE V. Consequences Grad v1 v2 v3 v4 v5

=0.102,0.137,0.129,0.086,0.121,0.102,0.092,0.129,0.102

C s as
T

C s = ARc B = 0.185

CONSEQUENCES GRADE DEFINITION Explanation The risks has almost no influence. Has economic loss of less RMB100,000 and no personnel casualty Economic loss of less RMB1000,000 and some personnel cacuslty Economic loss of less RMB10,000,000 and more personnel cacuslty Large ecomomic loss and forming heavy social and political influences

TABLE VI. Consequences u1 u2 u3 u4 u5 u6 u7 u8 v1 0 0 0.1 0.2 0 0 0.3 0

MEMBERSHIP DEGREE SET (RC) v2 0.4 0.3 0.4 0.2 0.2 0.3 0.3 0.4 v3 0.3 0.5 0.4 0.4 0.5 0.4 0.3 0.4 v4 0.2 0.1 0.1 0.2 0.2 0.2 0.1 0.1 v5 0.1 0.1 0 0 0.1 0.1 0 0.1

u9

0.1

0.4

0.3

0.2

4273

R can be calculated by formula (7).

R = Ps + C s Ps C s =0.275
2) Procedure (g) Use similar method analysis the risks of procedure (g), can get the following results.

Human errors identification in each work procedures and calculate probability of each errors with human factors model. Choosing the procedures that with high failure probability because of human errors to analyse the risks with entropy weight measurement method for determine the uncertainty in construction. Analyse the entropy weight coefficient results, determine the risk controlling keys. REFERENCES

A = (1 , 2 , 3 , 4 ) =0.340.1670.2950.198
Ps = AR p B T = 0.104 A = (1 , 2 , 3 , 4 ) =0.257,0.324,0.23,0.189

R = Ps + C s Ps C s =0.273
E. Risk Controlling Entropy weight coefficient shows that in procedure (e), human errors of wrong excavating method, exceeded excavating and exceeded load are relevant important. These errors are risk controlling keys. In procedure (f), material deviation is the main human errors that should be controlled. For procedure (g), errors of wrong excavating method and exceeded load have relevant larger entropy weight coefficient. So, in this procedure, risk controlled keys are these two kind of human errors. IV. CONCLUSIONS AND DISCUSSIONS In this research, a valid risk assessment method that considering human factors sufficiently has been provided for excavation engineering. This method includes three blocks for dynamic risk during the construction period.
[1]

C s = ARc B = 0.189

[2] [3]

[4]

[5]

[6] [7] [8]

Canto-Perello J., Curiel-Esparza J., Human factors engineering in utility tunnel design. Tunnelling and Underground Space Technology, 2001, 16, pp.211-215. Chapanis A., Research techniques in human engineering. Baltimore: Johns Hopkins Press, 1956, pp. 51-62. Abdelhamid T., Everette J., Physiological demands during construction work. Journal of Construction Engineering and Management, 2002, 128(5), pp.427-437. Vicknayson T.M.J., Perception of human risk factors in construction projects: an exploratory study. International Journal of Project Management, 2004, 22, pp.131-137. Cacciabue P. C., Human error risk management for engineering systems: a methodology for design, safety assessment, accident investigation and training. Reliability Engineering & System Safety, 2004, 83(2), pp.229-240. Shannon C.E., The mathematical theory of communication. Bell System Technical Journa, 1948, 27, pp. 379-423. Jaynes E.T., Information theory & statistical mechanics. The Physical Review, 1957, 166, pp. 620-630, & 108, pp. 171-190. Tah J.H.M., Carr V., A Proposal for Construction Project Risk Assessment Using Fuzzy Logic. Construction Management and Economics, 2000, 18(4), pp. 491-500.

4274