The prediction of overall consumption quantity. On the basis of effect factors such as security task, demand standard, security time, etc., scientific prediction is made on emergency material demanding quantity at some direction. Through the selection on different decision aims and the different calculation methods, the comprehensive decision results of emergency material demanding quantity is formed, and the predicted quantity of consumption is determined.

The determination of material demanding quantity in diverse security directions. Under the circumstance of emergency, the material security demand usually involves many directions, and the demanding quantities in different directions are not the same. The distance between every direction and security unit is also different [4D.B. Clarke, A. Chatterjee, and S.M. Rutner, "Intermodal freight transportation and highway safety", Transportation Quorterly, vol. A2, pp. 97-110, 1996.]. To realize the security optimization from multi-point to multi-point, firstly, we should base on the predicted consumption quantity as well as the emergency material demanding quantity in each direction.

The determination of security unit and its task volume distribution [5R.B. Copperman, M.P. Devlin, and R.M. Ewalt, "Coordinating and prioritizing multimodal transportation projects", In: Proceedings of Systems and Information Engineering Design Symposium, Charlottesvile, VA: USA, 2004, pp. 113-119.]. Based on emergency material security requests, relevant security unit is selected, and initial security task of each unit is determined, including the demanded quantity and kinds of materials. We used Dijkstra and its improving calculation to calculate road optimization, and got the security distance (road minimum) from each security unit to each security direction, so that to provide decision evidence of reserve distribution optimization.

Based on the initial security task, we can choose one from the three models. The model of building new reserve points, the model of adjusting reserve variety & quantity, and the model of adjusting security task, so that we can make the reserve distribution optimization [6A. Lozano, and G. Storchi, "Shortest viable path algorithm in multimodal networks", Transportation Research, vol. A35, pp. 225-241, 2001.].

Through the calculation of the shortest path of each security unit and various operational direction, if the shortest distance between all the security units and the operational directions is larger than the corresponding maximum security distance of the security unit, new reserves are required to be built; with the centroid method that is based on arbitrary coordinates, the closest centre point to all the node security distances in the network made up of nodes of each operational direction.

When the material reserve varieties and quantity of a selected security unit can not satisfy the demand of a task, adjustment of them should be made according to the results of security optimization calculation, by newly adding and dilatation approaches. The system selects the best adjusting object and the varieties and quantity that are required to adjust according to the calculation results of path optimization from multi-point to multi-point, as well as capacity size of each security unit [7J. Guo, Logistics Optimization, Scientific Publishing Company: Beijing, 2010.].

Fig. (1) Emergency material reserve distribution optimization procedure chart.

When the variety and quantity of material reserve of a selected security unit can satisfy the requirement of a task, through the analysis of path security optimization, in accordance with the principle of shortest time, shortest path and optimal efficiency, adjustment on the initial scheme is carried on to reallocation tasks of different security units. Also security scheme was made for departments’ decision-making. The procedure chart is shown in Fig. (1).

1. The demanded quantity w of each demanding point is the actual demand of the unit, and the total amount is known.
2. In the emergency material security, transportation fee can be assumed to be proportional to distance d.
3. The priority of demanding unit that needs to be secured can be represented by a weighted coefficient k.
4. The reserve point (x₀ , y₀ ) got from centroid method, is not an absolute point but relative. It is a scope.

We set each demanding point as A, B, C, D, E, F, G, H. A coordinate system is set, the original point is (0,0), shown in the chart. The coordinate of each demanding point is (x_i, y_i) i =1,2,3..; the coordinate of the selected reserve address is (x₀ , y₀ ), also the result we require; the transportation fee of each demanding point unit distance is p_i (i =1,2,3 …); d_iis the distance between each demanding point and original point(it can be non-straight line distance but road mileage). The sum of materials required by each demanding point (not considering the variety of material, but the variety has to meet every demand of security)is w_i(i = 1,2,...3); the security demanding factor of demanding point is K_i(i = 1,2,3 …); shown in Table 1.

Model 1: Only Considering Economic Efficiency, G₀ =19118281.

According to formula (4) and formula (5), we can get the vertical and horizontal coordinate of M:

x₀ = -55906472556/191182819=-292.5

y₀ = 40649787377/191182819=212.6

It is obvious that M (-292.5,212.6) is an obtained reserve point, shown as in Fig. (2).

Model 2: Considering Security and Economic Efficiency

According to formula (6) we can get:=1018721494.

According to formula (9) and formula (10) we can get the vertical and horizontal coordinate of N:

x₀ = 2.06933E+11/1018721494=-203

y₀ = -4.35519E+11/1018721494=428

It is obvious that N (-203,428) is another reserve point we obtained, shown in the following chart.

Fig. (2) The reserve point chart.