The Traffic Conflict Technique (TCT), as developed by General Motors
Research laboratories* was evaluated by the Federal Highway Administration
, in cooperation with the state highway departments of Washington, Ohio,
and Virginia. The counting of traffic conflicts allows an estimate to be
made of accident potential at specific locations, which include at-grade
intersections and road sections. Traffic accident records may ofen be dis
-torted, incorrect, incomplete, and not readily available.
Thus, methods such as TCT should be developed for identifying potential
accident sites.
A special field work plan was developed in complying with special street
conditions of Tainan, which provide special motorcycle lane and bicycle
lane beside motorcar lane. Fourteen intersections and twenty-one road
sections from C. B.D. of Tainan were selected as study locations. The data
was first analyzed with ratio of conflict to opportunity and accident to
conflict, then correlation between conflict and accident was examined with
rate, i.e., conflict or accident per vehicle. After significant correlation
between conflict and accident was identified, the mathematic models were
built by using backward elimination procedure.
The study has first aim of identitying the workability of TCT at Tainan.
The second aim of work is to establish the models for the prediction of
accidents. The last but not the least of the study purpose is to investigate
the application of TCT to traffic engineering.
The study was extended to the other mode, motorcycle, and then to road
sections. Although the study has its first success in this attempt,
further improvements are requited. For instance conflict detecting
standard for motorcycles should be modified to suit the free movement of
motorcycles. As for as road sections are concerned shorter section lengths
should be adopted instead of long links used in this study.
Finally for continuing study of TCT application in Tainan to Cove other
locations, in particular, suburb and rural areas are strongly recommended
for further studies.
美國通用汽車公司實驗所發展的交通衝突技術在聯邦公路總局和華盛頓、俄亥俄、維
吉尼亞等州公路局有良好的評價。交通衝突的測定可供估計特殊地點交通事故的潛在
可能性。這些地點包括十字路口及路段。因交通事故記錄往往不完整，曲解事實，而
失其可靠性，其資料系統不易蒐集。發展一種如交通衝突方式以資認別交通事故時常
發生地點而加以改善為目前急需要研究之問題。
對路況而言，台南市有別於美國，台南市的街道除了汽車快車道外，尚有機車道及腳
踏車慢車道。本研究先設計不同於美國的工作方案以測定交通衝突。於台南市中心區
選出14個交叉路口及21個路段，做研究地點。所獲得資料先按交通衝突與交通衝突機
會之比及交通事故與交通衝突之比，分別加以分析。然後分析交通衝突和交通事故間
之相關性。經證實兩者之間確有相關之後，使用往後遞減法求最佳數學模式。
本研究的研究目標為：第一、求交通衝突技術的可行性，其次、為設立預估交通事故
模式。最後，但很重要的研究目標為探求其應用。本研究將其研究對象推廣包括機車
及路段，雖然獲得初步成功，尚需繼續研究改善。例如機車的交通衝突測定標準稍加
以修正以符合其活動自由性。至於路段測定需縮短其觀察長度。
最後，繼續研究台南市衝突技術，將其研究範圍推廣至郊區及鄉下區以完成本研究是
最大的期望。