Estimation of the Effect of Rice Husk Ash on Portland Cement Setting Times with ANFIS

Eyyüp Gülbandılar; Burak Koçak; İzzet Özdemir; Yılmaz Koçak

Makale detayları

Estimation of the Effect of Rice Husk Ash on Portland Cement Setting Times with ANFIS

Yazarlar:

Eyyüp Gülbandılar | Burak Koçak | İzzet Özdemir | Yılmaz Koçak

Yayıncı:

Eskişehir Osmangazi University

Ana Başlıklar:

Bilgisayar Yazılımı

Dergi:

Eskişehir Türk Dünyası Uygulama ve Araştırma Merkezi Bilişim Dergisi Vol.4 Issue 2, pp.34-42

Belge Tipi:

Makale

Yayın Tarihi:

6/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001287660

pISSN:

2687-606X

Anahtar kelimeler:

Özet:

 In this study, setting times of rice husk-replaced cement pastes were estimated with an adaptive network-based fuzzy inference system ANFIS. In the first stage of the study, initial and final setting time of cement pastes in which 0, 2.5, 5, 7.5, 10, 12.5 and 15 rice husk ash substituted into Portland cement were determined. In the second stage, two models were created using the gaussian and triangular membership functions with the results obtained from the experiments, and with the help of these models, the initial and final setting times were tried to be estimated. In the last stage, three different statistical methods, namely R2, MAPE and RMSE, were used to understand the reliability of the results obtained from the models. According to the initial setting time obtained, R2, MAPE and RMSE were found as 0.9912, 0.0167, and 5.6537 for gaussian membership functions, 0.9957, 0.0119, and 3.9661 for triangle membership functions, respectively. According to the final setting time, R2, MAPE and RMSE were detected as 0.9909, 0.0144, and 5.8693 for gaussian membership functions, 0.9939, 0.0127, and 4.8125 for triangle membership functions, respectively. When the statistical indices were evaluated as a whole, it was determined that results very close to real values were obtained with the models created with both gaussian and triangular membership functions in terms of initial and final setting times. Therefore, it can be stated that both models are powerful models for estimating setting times.

 In this study, setting times of rice husk-replaced cement pastes were estimated with an adaptive network-based fuzzy inference system ANFIS. In the first stage of the study, initial and final setting time of cement pastes in which 0, 2.5, 5, 7.5, 10, 12.5 and 15 rice husk ash substituted into Portland cement were determined. In the second stage, two models were created using the gaussian and triangular membership functions with the results obtained from the experiments, and with the help of these models, the initial and final setting times were tried to be estimated. In the last stage, three different statistical methods, namely R2, MAPE and RMSE, were used to understand the reliability of the results obtained from the models. According to the initial setting time obtained, R2, MAPE and RMSE were found as 0.9912, 0.0167, and 5.6537 for gaussian membership functions, 0.9957, 0.0119, and 3.9661 for triangle membership functions, respectively. According to the final setting time, R2, MAPE and RMSE were detected as 0.9909, 0.0144, and 5.8693 for gaussian membership functions, 0.9939, 0.0127, and 4.8125 for triangle membership functions, respectively. When the statistical indices were evaluated as a whole, it was determined that results very close to real values were obtained with the models created with both gaussian and triangular membership functions in terms of initial and final setting times. Therefore, it can be stated that both models are powerful models for estimating setting times.

 In this study, setting times of rice husk-replaced cement pastes were estimated with an adaptive network-based fuzzy inference system ANFIS. In the first stage of the study, initial and final setting time of cement pastes in which 0, 2.5, 5, 7.5, 10, 12.5 and 15 rice husk ash substituted into Portland cement were determined. In the second stage, two models were created using the gaussian and triangular membership functions with the results obtained from the experiments, and with the help of these models, the initial and final setting times were tried to be estimated. In the last stage, three different statistical methods, namely R2, MAPE and RMSE, were used to understand the reliability of the results obtained from the models. According to the initial setting time obtained, R2, MAPE and RMSE were found as 0.9912, 0.0167, and 5.6537 for gaussian membership functions, 0.9957, 0.0119, and 3.9661 for triangle membership functions, respectively. According to the final setting time, R2, MAPE and RMSE were detected as 0.9909, 0.0144, and 5.8693 for gaussian membership functions, 0.9939, 0.0127, and 4.8125 for triangle membership functions, respectively. When the statistical indices were evaluated as a whole, it was determined that results very close to real values were obtained with the models created with both gaussian and triangular membership functions in terms of initial and final setting times. Therefore, it can be stated that both models are powerful models for estimating setting times.

 Bu çalışmada, pirinç kabuğu ikameli çimento hamurlarının priz süreleri uyarlamalı ağ tabanlı bulanık çıkarım sistemi ANFIS ile tahmin edilmiştir. Çalışmanın ilk aşamasında, Portland çimentosu içerisine 0, 2.5, 5, 7.5, 10, 12.5 ve 15 oranlarında pirinç kabuğu külü ikame edilmiş olan çimento hamurlarının priz başlangıç ve priz sonu süreleri belirlenmiştir. İkinci aşamada, deneylerden elde edilen sonuçlarla gauss ve üçgen üyelik fonksiyonları kullanılarak iki model oluşturulmuş ve bu modeller yardımıyla priz başlangıç ve priz sonu süreleri tahmin edilmeye çalışılmıştır. Son aşamada ise modellerden elde edilen sonuçlarının güvenilirliği anlayabilmek için R2, MAPE ve RMSE olmak üzere üç farklı istatiksel yöntem kullanılmıştır. Elde edilen priz başlangıç sürelerine göre R2, MAPE ve RMSE değerleri sırasıyla; gauss üyelik fonksiyonları için 0.9912, 0.0167 ve 5.6537, üçgen üyelik fonksiyonları için ise 0.9957, 0.0119 ve 3.9661 olarak bulunmuştur. Priz sonu sürelerine göre ise R2, MAPE ve RMSE değerleri sırasıyla; gauss üyelik fonksiyonları için 0.9909, 0.0144 ve 5.8693, üçgen üyelik fonksiyonları için 0.9939, 0.0127 ve 4.8125 olarak tespit edilmiştir. İstatistiksel endeksler bir bütün olarak değerlendirildiğinde, priz başlangıç ve priz sonu süreleri açısından hem gauss hem de üçgen üyelik fonksiyonlarıyla oluşturulan modellerle gerçek değerlere çok yakın sonuçlar elde edildiği belirlenmiştir. Dolayısı ile her iki modelin de priz sürelerini tahmin etmek için güçlü modeller olduğu ifade edilebilir.

Makale detayları

Estimation of the Effect of Rice Husk Ash on Portland Cement Setting Times with ANFIS

Yazarlar:

Eyyüp Gülbandılar | Burak Koçak | İzzet Özdemir | Yılmaz Koçak

Yayıncı:

Eskişehir Osmangazi University

Ana Başlıklar:

Bilgisayar Yazılımı

Dergi:

Eskişehir Türk Dünyası Uygulama ve Araştırma Merkezi Bilişim Dergisi Vol.4 Issue 2, pp.34-42

Belge Tipi:

Makale

Yayın Tarihi:

6/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001287660

pISSN:

2687-606X

ANFIS

Portland Cement

Rice husk ash

Setting timeANFIS

Portland çimento

Pirinç kabuğu külü

Priz süresi

Sayı

Makale detayları

Yazarlar: Eyyüp Gülbandılar | Burak Koçak | İzzet Özdemir | Yılmaz Koçak

Yayıncı: Eskişehir Osmangazi University

Ana Başlıklar: Bilgisayar Yazılımı

Dergi: Eskişehir Türk Dünyası Uygulama ve Araştırma Merkezi Bilişim Dergisi Vol.4 Issue 2, pp.34-42

Belge Tipi: Makale

Yayın Tarihi: 6/2023

Belge Dili: Türkçe

Document ID (Org No.): 2001287660

pISSN: 2687-606X

Sayı

Yazarlar:

Eyyüp Gülbandılar | Burak Koçak | İzzet Özdemir | Yılmaz Koçak

Yayıncı:

Eskişehir Osmangazi University

Ana Başlıklar:

Bilgisayar Yazılımı

Dergi:

Eskişehir Türk Dünyası Uygulama ve Araştırma Merkezi Bilişim Dergisi Vol.4 Issue 2, pp.34-42

Belge Tipi:

Makale

Yayın Tarihi:

6/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001287660

pISSN:

2687-606X